Three issues flagged by the Copilot review on this PR:

1. Double JSON emit on stage failure (Copilot #1, #2). When -Stage <name>
   ran a worker that threw, Invoke-Stage's finally emitted a JSON result
   frame AND the entry-point catch emitted a second error frame --
   producing two concatenated JSON objects on stdout and breaking the
   one-line-per-invocation contract that drivers parse against. Same
   issue applied to -Json mode on a full install (every stage's finally
   plus a final error frame missing duration_ms/skipped).

   Fix: Invoke-Stage's finally now sets $script:_StageEmittedErrorFrame
   when it emits a failure frame; the entry-point catch checks the flag
   and skips its own emit, still exit 1.

2. $prevEAP uninitialized on early try-block throw (Copilot #3). In
   Install-Uv, Test-Python, Test-Node's winget fallback,
   _Run-NpmInstall, and the playwright block, '$prevEAP =
   $ErrorActionPreference' lived as the first statement INSIDE the
   try. If anything between 'try {' and that line threw (Write-Info on
   an unusual host, the npx-finding loop, etc.), the catch's
   'if ($prevEAP) { ... }' restore was a no-op and EAP could remain
   relaxed.

   Fix: hoist '$prevEAP = $ErrorActionPreference' to the line
   immediately before 'try {' in all five sites. Catch's restore is
   now always meaningful regardless of where in the try the throw
   originated.

No change to Invoke-Stage's success path or to the four lint-clean EAP
sites (Test-Node was the only winget-related catch). All 19 metadata
smoke tests still pass.

… ops

Three install.ps1 improvements pulled from the thin-installer work on
bb/gui (PR NousResearch#27822) that benefit the canonical CLI install flow on main:

1. Strip UTF-8 BOM from scripts/install.ps1.

   The canonical 'irm <raw URL> | iex' install flow has been broken
   since commit 4279da4 re-introduced a UTF-8 BOM that PR NousResearch#27224
   had explicitly stripped. PowerShell 5.1's 'irm' returns the
   response body as a string with the BOM surviving as a leading
   \ufeff character; 'iex' then evaluates that string and the parser
   chokes on the invisible character before param(), surfacing as a
   cascade of 'The assignment expression is not valid' errors at
   every param default value.

   File body is verified pure ASCII (no character above byte 127),
   so PS 5.1 with no BOM falls back to Windows-1252 decoding which
   is identical to ASCII for our content. Both install paths work:
     - 'irm ... | iex' (canonical one-liner)
     - 'powershell -File install.ps1' (programmatic / desktop bootstrap)

2. New -Commit and -Tag string params for reproducible pinning.

   Higher-precedence variants of -Branch. When set, the repository
   stage clones $Branch (fast partial fetch) and then 'git checkout's
   the exact ref. Precedence: Commit > Tag > Branch. Honoured by all
   three code paths:
     - Update path (existing valid checkout): fetch + checkout
       --detach <commit|tag> instead of checkout + pull.
     - Fresh clone: clone --branch $Branch, then post-clone
       'git checkout --detach' to the requested ref.
     - ZIP fallback: pick archive URL for the most-specific ref
       (commit -> archive/<sha>.zip, tag -> archive/refs/tags/
       <tag>.zip, else archive/refs/heads/<branch>.zip).

   Used by the Hermes desktop's first-launch bootstrap to pin the
   .exe to the exact commit it was built against, so the cloned
   Hermes Agent tree always matches what the .exe was tested with.
   Also enables release-bundle pinning (e.g. Microsoft Store builds
   pinning to a release tag) and CI reproducibility.

3. EAP=Continue wrap around the new pin-step git invocations.

   'git fetch origin <commit>' writes the routine 'From <url>' info
   line to stderr. Under the script's global $ErrorActionPreference
   = 'Stop' that stderr line is wrapped as an ErrorRecord and
   terminates the script even though fetch+checkout actually succeed.
   Same EAP=Stop + native-stderr footgun we hit during the install.ps1
   hardening pass in Install-Uv, Test-Python, _Run-NpmInstall.

   Wrap both the update-path fetch/checkout block AND the post-clone
   pin block in $ErrorActionPreference = 'Continue' (restored in
   finally). Real failures still caught by $LASTEXITCODE checks.

Three issues flagged by the Copilot review on this PR:

1. Double JSON emit on stage failure (Copilot #1, #2). When -Stage <name>
   ran a worker that threw, Invoke-Stage's finally emitted a JSON result
   frame AND the entry-point catch emitted a second error frame --
   producing two concatenated JSON objects on stdout and breaking the
   one-line-per-invocation contract that drivers parse against. Same
   issue applied to -Json mode on a full install (every stage's finally
   plus a final error frame missing duration_ms/skipped).

   Fix: Invoke-Stage's finally now sets $script:_StageEmittedErrorFrame
   when it emits a failure frame; the entry-point catch checks the flag
   and skips its own emit, still exit 1.

2. $prevEAP uninitialized on early try-block throw (Copilot #3). In
   Install-Uv, Test-Python, Test-Node's winget fallback,
   _Run-NpmInstall, and the playwright block, '$prevEAP =
   $ErrorActionPreference' lived as the first statement INSIDE the
   try. If anything between 'try {' and that line threw (Write-Info on
   an unusual host, the npx-finding loop, etc.), the catch's
   'if ($prevEAP) { ... }' restore was a no-op and EAP could remain
   relaxed.

   Fix: hoist '$prevEAP = $ErrorActionPreference' to the line
   immediately before 'try {' in all five sites. Catch's restore is
   now always meaningful regardless of where in the try the throw
   originated.

No change to Invoke-Stage's success path or to the four lint-clean EAP
sites (Test-Node was the only winget-related catch). All 19 metadata
smoke tests still pass.

… ops

Three install.ps1 improvements pulled from the thin-installer work on
bb/gui (PR NousResearch#27822) that benefit the canonical CLI install flow on main:

1. Strip UTF-8 BOM from scripts/install.ps1.

   The canonical 'irm <raw URL> | iex' install flow has been broken
   since commit 4279da4 re-introduced a UTF-8 BOM that PR NousResearch#27224
   had explicitly stripped. PowerShell 5.1's 'irm' returns the
   response body as a string with the BOM surviving as a leading
   \ufeff character; 'iex' then evaluates that string and the parser
   chokes on the invisible character before param(), surfacing as a
   cascade of 'The assignment expression is not valid' errors at
   every param default value.

   File body is verified pure ASCII (no character above byte 127),
   so PS 5.1 with no BOM falls back to Windows-1252 decoding which
   is identical to ASCII for our content. Both install paths work:
     - 'irm ... | iex' (canonical one-liner)
     - 'powershell -File install.ps1' (programmatic / desktop bootstrap)

2. New -Commit and -Tag string params for reproducible pinning.

   Higher-precedence variants of -Branch. When set, the repository
   stage clones $Branch (fast partial fetch) and then 'git checkout's
   the exact ref. Precedence: Commit > Tag > Branch. Honoured by all
   three code paths:
     - Update path (existing valid checkout): fetch + checkout
       --detach <commit|tag> instead of checkout + pull.
     - Fresh clone: clone --branch $Branch, then post-clone
       'git checkout --detach' to the requested ref.
     - ZIP fallback: pick archive URL for the most-specific ref
       (commit -> archive/<sha>.zip, tag -> archive/refs/tags/
       <tag>.zip, else archive/refs/heads/<branch>.zip).

   Used by the Hermes desktop's first-launch bootstrap to pin the
   .exe to the exact commit it was built against, so the cloned
   Hermes Agent tree always matches what the .exe was tested with.
   Also enables release-bundle pinning (e.g. Microsoft Store builds
   pinning to a release tag) and CI reproducibility.

3. EAP=Continue wrap around the new pin-step git invocations.

   'git fetch origin <commit>' writes the routine 'From <url>' info
   line to stderr. Under the script's global $ErrorActionPreference
   = 'Stop' that stderr line is wrapped as an ErrorRecord and
   terminates the script even though fetch+checkout actually succeed.
   Same EAP=Stop + native-stderr footgun we hit during the install.ps1
   hardening pass in Install-Uv, Test-Python, _Run-NpmInstall.

   Wrap both the update-path fetch/checkout block AND the post-clone
   pin block in $ErrorActionPreference = 'Continue' (restored in
   finally). Real failures still caught by $LASTEXITCODE checks.

Three issues flagged by the Copilot review on this PR:

1. Double JSON emit on stage failure (Copilot #1, #2). When -Stage <name>
   ran a worker that threw, Invoke-Stage's finally emitted a JSON result
   frame AND the entry-point catch emitted a second error frame --
   producing two concatenated JSON objects on stdout and breaking the
   one-line-per-invocation contract that drivers parse against. Same
   issue applied to -Json mode on a full install (every stage's finally
   plus a final error frame missing duration_ms/skipped).

   Fix: Invoke-Stage's finally now sets $script:_StageEmittedErrorFrame
   when it emits a failure frame; the entry-point catch checks the flag
   and skips its own emit, still exit 1.

2. $prevEAP uninitialized on early try-block throw (Copilot #3). In
   Install-Uv, Test-Python, Test-Node's winget fallback,
   _Run-NpmInstall, and the playwright block, '$prevEAP =
   $ErrorActionPreference' lived as the first statement INSIDE the
   try. If anything between 'try {' and that line threw (Write-Info on
   an unusual host, the npx-finding loop, etc.), the catch's
   'if ($prevEAP) { ... }' restore was a no-op and EAP could remain
   relaxed.

   Fix: hoist '$prevEAP = $ErrorActionPreference' to the line
   immediately before 'try {' in all five sites. Catch's restore is
   now always meaningful regardless of where in the try the throw
   originated.

No change to Invoke-Stage's success path or to the four lint-clean EAP
sites (Test-Node was the only winget-related catch). All 19 metadata
smoke tests still pass.

… ops

Three install.ps1 improvements pulled from the thin-installer work on
bb/gui (PR #27822) that benefit the canonical CLI install flow on main:

1. Strip UTF-8 BOM from scripts/install.ps1.

   The canonical 'irm <raw URL> | iex' install flow has been broken
   since commit dec2556 re-introduced a UTF-8 BOM that PR #27224
   had explicitly stripped. PowerShell 5.1's 'irm' returns the
   response body as a string with the BOM surviving as a leading
   \ufeff character; 'iex' then evaluates that string and the parser
   chokes on the invisible character before param(), surfacing as a
   cascade of 'The assignment expression is not valid' errors at
   every param default value.

   File body is verified pure ASCII (no character above byte 127),
   so PS 5.1 with no BOM falls back to Windows-1252 decoding which
   is identical to ASCII for our content. Both install paths work:
     - 'irm ... | iex' (canonical one-liner)
     - 'powershell -File install.ps1' (programmatic / desktop bootstrap)

2. New -Commit and -Tag string params for reproducible pinning.

   Higher-precedence variants of -Branch. When set, the repository
   stage clones $Branch (fast partial fetch) and then 'git checkout's
   the exact ref. Precedence: Commit > Tag > Branch. Honoured by all
   three code paths:
     - Update path (existing valid checkout): fetch + checkout
       --detach <commit|tag> instead of checkout + pull.
     - Fresh clone: clone --branch $Branch, then post-clone
       'git checkout --detach' to the requested ref.
     - ZIP fallback: pick archive URL for the most-specific ref
       (commit -> archive/<sha>.zip, tag -> archive/refs/tags/
       <tag>.zip, else archive/refs/heads/<branch>.zip).

   Used by the Hermes desktop's first-launch bootstrap to pin the
   .exe to the exact commit it was built against, so the cloned
   Hermes Agent tree always matches what the .exe was tested with.
   Also enables release-bundle pinning (e.g. Microsoft Store builds
   pinning to a release tag) and CI reproducibility.

3. EAP=Continue wrap around the new pin-step git invocations.

   'git fetch origin <commit>' writes the routine 'From <url>' info
   line to stderr. Under the script's global $ErrorActionPreference
   = 'Stop' that stderr line is wrapped as an ErrorRecord and
   terminates the script even though fetch+checkout actually succeed.
   Same EAP=Stop + native-stderr footgun we hit during the install.ps1
   hardening pass in Install-Uv, Test-Python, _Run-NpmInstall.

   Wrap both the update-path fetch/checkout block AND the post-clone
   pin block in $ErrorActionPreference = 'Continue' (restored in
   finally). Real failures still caught by $LASTEXITCODE checks.

Three issues flagged by the Copilot review on this PR:

1. Double JSON emit on stage failure (Copilot NousResearch#1, NousResearch#2). When -Stage <name>
   ran a worker that threw, Invoke-Stage's finally emitted a JSON result
   frame AND the entry-point catch emitted a second error frame --
   producing two concatenated JSON objects on stdout and breaking the
   one-line-per-invocation contract that drivers parse against. Same
   issue applied to -Json mode on a full install (every stage's finally
   plus a final error frame missing duration_ms/skipped).

   Fix: Invoke-Stage's finally now sets $script:_StageEmittedErrorFrame
   when it emits a failure frame; the entry-point catch checks the flag
   and skips its own emit, still exit 1.

2. $prevEAP uninitialized on early try-block throw (Copilot NousResearch#3). In
   Install-Uv, Test-Python, Test-Node's winget fallback,
   _Run-NpmInstall, and the playwright block, '$prevEAP =
   $ErrorActionPreference' lived as the first statement INSIDE the
   try. If anything between 'try {' and that line threw (Write-Info on
   an unusual host, the npx-finding loop, etc.), the catch's
   'if ($prevEAP) { ... }' restore was a no-op and EAP could remain
   relaxed.

   Fix: hoist '$prevEAP = $ErrorActionPreference' to the line
   immediately before 'try {' in all five sites. Catch's restore is
   now always meaningful regardless of where in the try the throw
   originated.

No change to Invoke-Stage's success path or to the four lint-clean EAP
sites (Test-Node was the only winget-related catch). All 19 metadata
smoke tests still pass.

… ops

Three install.ps1 improvements pulled from the thin-installer work on
bb/gui (PR NousResearch#27822) that benefit the canonical CLI install flow on main:

1. Strip UTF-8 BOM from scripts/install.ps1.

   The canonical 'irm <raw URL> | iex' install flow has been broken
   since commit 2089537 re-introduced a UTF-8 BOM that PR NousResearch#27224
   had explicitly stripped. PowerShell 5.1's 'irm' returns the
   response body as a string with the BOM surviving as a leading
   \ufeff character; 'iex' then evaluates that string and the parser
   chokes on the invisible character before param(), surfacing as a
   cascade of 'The assignment expression is not valid' errors at
   every param default value.

   File body is verified pure ASCII (no character above byte 127),
   so PS 5.1 with no BOM falls back to Windows-1252 decoding which
   is identical to ASCII for our content. Both install paths work:
     - 'irm ... | iex' (canonical one-liner)
     - 'powershell -File install.ps1' (programmatic / desktop bootstrap)

2. New -Commit and -Tag string params for reproducible pinning.

   Higher-precedence variants of -Branch. When set, the repository
   stage clones $Branch (fast partial fetch) and then 'git checkout's
   the exact ref. Precedence: Commit > Tag > Branch. Honoured by all
   three code paths:
     - Update path (existing valid checkout): fetch + checkout
       --detach <commit|tag> instead of checkout + pull.
     - Fresh clone: clone --branch $Branch, then post-clone
       'git checkout --detach' to the requested ref.
     - ZIP fallback: pick archive URL for the most-specific ref
       (commit -> archive/<sha>.zip, tag -> archive/refs/tags/
       <tag>.zip, else archive/refs/heads/<branch>.zip).

   Used by the Hermes desktop's first-launch bootstrap to pin the
   .exe to the exact commit it was built against, so the cloned
   Hermes Agent tree always matches what the .exe was tested with.
   Also enables release-bundle pinning (e.g. Microsoft Store builds
   pinning to a release tag) and CI reproducibility.

3. EAP=Continue wrap around the new pin-step git invocations.

   'git fetch origin <commit>' writes the routine 'From <url>' info
   line to stderr. Under the script's global $ErrorActionPreference
   = 'Stop' that stderr line is wrapped as an ErrorRecord and
   terminates the script even though fetch+checkout actually succeed.
   Same EAP=Stop + native-stderr footgun we hit during the install.ps1
   hardening pass in Install-Uv, Test-Python, _Run-NpmInstall.

   Wrap both the update-path fetch/checkout block AND the post-clone
   pin block in $ErrorActionPreference = 'Continue' (restored in
   finally). Real failures still caught by $LASTEXITCODE checks.

Three issues flagged by the Copilot review on this PR:

1. Double JSON emit on stage failure (Copilot #1, #2). When -Stage <name>
   ran a worker that threw, Invoke-Stage's finally emitted a JSON result
   frame AND the entry-point catch emitted a second error frame --
   producing two concatenated JSON objects on stdout and breaking the
   one-line-per-invocation contract that drivers parse against. Same
   issue applied to -Json mode on a full install (every stage's finally
   plus a final error frame missing duration_ms/skipped).

   Fix: Invoke-Stage's finally now sets $script:_StageEmittedErrorFrame
   when it emits a failure frame; the entry-point catch checks the flag
   and skips its own emit, still exit 1.

2. $prevEAP uninitialized on early try-block throw (Copilot #3). In
   Install-Uv, Test-Python, Test-Node's winget fallback,
   _Run-NpmInstall, and the playwright block, '$prevEAP =
   $ErrorActionPreference' lived as the first statement INSIDE the
   try. If anything between 'try {' and that line threw (Write-Info on
   an unusual host, the npx-finding loop, etc.), the catch's
   'if ($prevEAP) { ... }' restore was a no-op and EAP could remain
   relaxed.

   Fix: hoist '$prevEAP = $ErrorActionPreference' to the line
   immediately before 'try {' in all five sites. Catch's restore is
   now always meaningful regardless of where in the try the throw
   originated.

No change to Invoke-Stage's success path or to the four lint-clean EAP
sites (Test-Node was the only winget-related catch). All 19 metadata
smoke tests still pass.

… ops

Three install.ps1 improvements pulled from the thin-installer work on
bb/gui (PR NousResearch#27822) that benefit the canonical CLI install flow on main:

1. Strip UTF-8 BOM from scripts/install.ps1.

   The canonical 'irm <raw URL> | iex' install flow has been broken
   since commit dfc218d re-introduced a UTF-8 BOM that PR NousResearch#27224
   had explicitly stripped. PowerShell 5.1's 'irm' returns the
   response body as a string with the BOM surviving as a leading
   \ufeff character; 'iex' then evaluates that string and the parser
   chokes on the invisible character before param(), surfacing as a
   cascade of 'The assignment expression is not valid' errors at
   every param default value.

   File body is verified pure ASCII (no character above byte 127),
   so PS 5.1 with no BOM falls back to Windows-1252 decoding which
   is identical to ASCII for our content. Both install paths work:
     - 'irm ... | iex' (canonical one-liner)
     - 'powershell -File install.ps1' (programmatic / desktop bootstrap)

2. New -Commit and -Tag string params for reproducible pinning.

   Higher-precedence variants of -Branch. When set, the repository
   stage clones $Branch (fast partial fetch) and then 'git checkout's
   the exact ref. Precedence: Commit > Tag > Branch. Honoured by all
   three code paths:
     - Update path (existing valid checkout): fetch + checkout
       --detach <commit|tag> instead of checkout + pull.
     - Fresh clone: clone --branch $Branch, then post-clone
       'git checkout --detach' to the requested ref.
     - ZIP fallback: pick archive URL for the most-specific ref
       (commit -> archive/<sha>.zip, tag -> archive/refs/tags/
       <tag>.zip, else archive/refs/heads/<branch>.zip).

   Used by the Hermes desktop's first-launch bootstrap to pin the
   .exe to the exact commit it was built against, so the cloned
   Hermes Agent tree always matches what the .exe was tested with.
   Also enables release-bundle pinning (e.g. Microsoft Store builds
   pinning to a release tag) and CI reproducibility.

3. EAP=Continue wrap around the new pin-step git invocations.

   'git fetch origin <commit>' writes the routine 'From <url>' info
   line to stderr. Under the script's global $ErrorActionPreference
   = 'Stop' that stderr line is wrapped as an ErrorRecord and
   terminates the script even though fetch+checkout actually succeed.
   Same EAP=Stop + native-stderr footgun we hit during the install.ps1
   hardening pass in Install-Uv, Test-Python, _Run-NpmInstall.

   Wrap both the update-path fetch/checkout block AND the post-clone
   pin block in $ErrorActionPreference = 'Continue' (restored in
   finally). Real failures still caught by $LASTEXITCODE checks.

Three issues flagged by the Copilot review on this PR:

1. Double JSON emit on stage failure (Copilot #1, #2). When -Stage <name>
   ran a worker that threw, Invoke-Stage's finally emitted a JSON result
   frame AND the entry-point catch emitted a second error frame --
   producing two concatenated JSON objects on stdout and breaking the
   one-line-per-invocation contract that drivers parse against. Same
   issue applied to -Json mode on a full install (every stage's finally
   plus a final error frame missing duration_ms/skipped).

   Fix: Invoke-Stage's finally now sets $script:_StageEmittedErrorFrame
   when it emits a failure frame; the entry-point catch checks the flag
   and skips its own emit, still exit 1.

2. $prevEAP uninitialized on early try-block throw (Copilot #3). In
   Install-Uv, Test-Python, Test-Node's winget fallback,
   _Run-NpmInstall, and the playwright block, '$prevEAP =
   $ErrorActionPreference' lived as the first statement INSIDE the
   try. If anything between 'try {' and that line threw (Write-Info on
   an unusual host, the npx-finding loop, etc.), the catch's
   'if ($prevEAP) { ... }' restore was a no-op and EAP could remain
   relaxed.

   Fix: hoist '$prevEAP = $ErrorActionPreference' to the line
   immediately before 'try {' in all five sites. Catch's restore is
   now always meaningful regardless of where in the try the throw
   originated.

No change to Invoke-Stage's success path or to the four lint-clean EAP
sites (Test-Node was the only winget-related catch). All 19 metadata
smoke tests still pass.

… ops

Three install.ps1 improvements pulled from the thin-installer work on
bb/gui (PR NousResearch#27822) that benefit the canonical CLI install flow on main:

1. Strip UTF-8 BOM from scripts/install.ps1.

   The canonical 'irm <raw URL> | iex' install flow has been broken
   since commit dfc218d re-introduced a UTF-8 BOM that PR NousResearch#27224
   had explicitly stripped. PowerShell 5.1's 'irm' returns the
   response body as a string with the BOM surviving as a leading
   \ufeff character; 'iex' then evaluates that string and the parser
   chokes on the invisible character before param(), surfacing as a
   cascade of 'The assignment expression is not valid' errors at
   every param default value.

   File body is verified pure ASCII (no character above byte 127),
   so PS 5.1 with no BOM falls back to Windows-1252 decoding which
   is identical to ASCII for our content. Both install paths work:
     - 'irm ... | iex' (canonical one-liner)
     - 'powershell -File install.ps1' (programmatic / desktop bootstrap)

2. New -Commit and -Tag string params for reproducible pinning.

   Higher-precedence variants of -Branch. When set, the repository
   stage clones $Branch (fast partial fetch) and then 'git checkout's
   the exact ref. Precedence: Commit > Tag > Branch. Honoured by all
   three code paths:
     - Update path (existing valid checkout): fetch + checkout
       --detach <commit|tag> instead of checkout + pull.
     - Fresh clone: clone --branch $Branch, then post-clone
       'git checkout --detach' to the requested ref.
     - ZIP fallback: pick archive URL for the most-specific ref
       (commit -> archive/<sha>.zip, tag -> archive/refs/tags/
       <tag>.zip, else archive/refs/heads/<branch>.zip).

   Used by the Hermes desktop's first-launch bootstrap to pin the
   .exe to the exact commit it was built against, so the cloned
   Hermes Agent tree always matches what the .exe was tested with.
   Also enables release-bundle pinning (e.g. Microsoft Store builds
   pinning to a release tag) and CI reproducibility.

3. EAP=Continue wrap around the new pin-step git invocations.

   'git fetch origin <commit>' writes the routine 'From <url>' info
   line to stderr. Under the script's global $ErrorActionPreference
   = 'Stop' that stderr line is wrapped as an ErrorRecord and
   terminates the script even though fetch+checkout actually succeed.
   Same EAP=Stop + native-stderr footgun we hit during the install.ps1
   hardening pass in Install-Uv, Test-Python, _Run-NpmInstall.

   Wrap both the update-path fetch/checkout block AND the post-clone
   pin block in $ErrorActionPreference = 'Continue' (restored in
   finally). Real failures still caught by $LASTEXITCODE checks.

Three issues flagged by the Copilot review on this PR:

1. Double JSON emit on stage failure (Copilot #1, #2). When -Stage <name>
   ran a worker that threw, Invoke-Stage's finally emitted a JSON result
   frame AND the entry-point catch emitted a second error frame --
   producing two concatenated JSON objects on stdout and breaking the
   one-line-per-invocation contract that drivers parse against. Same
   issue applied to -Json mode on a full install (every stage's finally
   plus a final error frame missing duration_ms/skipped).

   Fix: Invoke-Stage's finally now sets $script:_StageEmittedErrorFrame
   when it emits a failure frame; the entry-point catch checks the flag
   and skips its own emit, still exit 1.

2. $prevEAP uninitialized on early try-block throw (Copilot #3). In
   Install-Uv, Test-Python, Test-Node's winget fallback,
   _Run-NpmInstall, and the playwright block, '$prevEAP =
   $ErrorActionPreference' lived as the first statement INSIDE the
   try. If anything between 'try {' and that line threw (Write-Info on
   an unusual host, the npx-finding loop, etc.), the catch's
   'if ($prevEAP) { ... }' restore was a no-op and EAP could remain
   relaxed.

   Fix: hoist '$prevEAP = $ErrorActionPreference' to the line
   immediately before 'try {' in all five sites. Catch's restore is
   now always meaningful regardless of where in the try the throw
   originated.

No change to Invoke-Stage's success path or to the four lint-clean EAP
sites (Test-Node was the only winget-related catch). All 19 metadata
smoke tests still pass.

… ops

Three install.ps1 improvements pulled from the thin-installer work on
bb/gui (PR NousResearch#27822) that benefit the canonical CLI install flow on main:

1. Strip UTF-8 BOM from scripts/install.ps1.

   The canonical 'irm <raw URL> | iex' install flow has been broken
   since commit 4279da4 re-introduced a UTF-8 BOM that PR NousResearch#27224
   had explicitly stripped. PowerShell 5.1's 'irm' returns the
   response body as a string with the BOM surviving as a leading
   \ufeff character; 'iex' then evaluates that string and the parser
   chokes on the invisible character before param(), surfacing as a
   cascade of 'The assignment expression is not valid' errors at
   every param default value.

   File body is verified pure ASCII (no character above byte 127),
   so PS 5.1 with no BOM falls back to Windows-1252 decoding which
   is identical to ASCII for our content. Both install paths work:
     - 'irm ... | iex' (canonical one-liner)
     - 'powershell -File install.ps1' (programmatic / desktop bootstrap)

2. New -Commit and -Tag string params for reproducible pinning.

   Higher-precedence variants of -Branch. When set, the repository
   stage clones $Branch (fast partial fetch) and then 'git checkout's
   the exact ref. Precedence: Commit > Tag > Branch. Honoured by all
   three code paths:
     - Update path (existing valid checkout): fetch + checkout
       --detach <commit|tag> instead of checkout + pull.
     - Fresh clone: clone --branch $Branch, then post-clone
       'git checkout --detach' to the requested ref.
     - ZIP fallback: pick archive URL for the most-specific ref
       (commit -> archive/<sha>.zip, tag -> archive/refs/tags/
       <tag>.zip, else archive/refs/heads/<branch>.zip).

   Used by the Hermes desktop's first-launch bootstrap to pin the
   .exe to the exact commit it was built against, so the cloned
   Hermes Agent tree always matches what the .exe was tested with.
   Also enables release-bundle pinning (e.g. Microsoft Store builds
   pinning to a release tag) and CI reproducibility.

3. EAP=Continue wrap around the new pin-step git invocations.

   'git fetch origin <commit>' writes the routine 'From <url>' info
   line to stderr. Under the script's global $ErrorActionPreference
   = 'Stop' that stderr line is wrapped as an ErrorRecord and
   terminates the script even though fetch+checkout actually succeed.
   Same EAP=Stop + native-stderr footgun we hit during the install.ps1
   hardening pass in Install-Uv, Test-Python, _Run-NpmInstall.

   Wrap both the update-path fetch/checkout block AND the post-clone
   pin block in $ErrorActionPreference = 'Continue' (restored in
   finally). Real failures still caught by $LASTEXITCODE checks.

Three issues flagged by the Copilot review on this PR:

1. Double JSON emit on stage failure (Copilot NousResearch#1, NousResearch#2). When -Stage <name>
   ran a worker that threw, Invoke-Stage's finally emitted a JSON result
   frame AND the entry-point catch emitted a second error frame --
   producing two concatenated JSON objects on stdout and breaking the
   one-line-per-invocation contract that drivers parse against. Same
   issue applied to -Json mode on a full install (every stage's finally
   plus a final error frame missing duration_ms/skipped).

   Fix: Invoke-Stage's finally now sets $script:_StageEmittedErrorFrame
   when it emits a failure frame; the entry-point catch checks the flag
   and skips its own emit, still exit 1.

2. $prevEAP uninitialized on early try-block throw (Copilot NousResearch#3). In
   Install-Uv, Test-Python, Test-Node's winget fallback,
   _Run-NpmInstall, and the playwright block, '$prevEAP =
   $ErrorActionPreference' lived as the first statement INSIDE the
   try. If anything between 'try {' and that line threw (Write-Info on
   an unusual host, the npx-finding loop, etc.), the catch's
   'if ($prevEAP) { ... }' restore was a no-op and EAP could remain
   relaxed.

   Fix: hoist '$prevEAP = $ErrorActionPreference' to the line
   immediately before 'try {' in all five sites. Catch's restore is
   now always meaningful regardless of where in the try the throw
   originated.

No change to Invoke-Stage's success path or to the four lint-clean EAP
sites (Test-Node was the only winget-related catch). All 19 metadata
smoke tests still pass.

… ops

Three install.ps1 improvements pulled from the thin-installer work on
bb/gui (PR NousResearch#27822) that benefit the canonical CLI install flow on main:

1. Strip UTF-8 BOM from scripts/install.ps1.

   The canonical 'irm <raw URL> | iex' install flow has been broken
   since commit 4279da4 re-introduced a UTF-8 BOM that PR NousResearch#27224
   had explicitly stripped. PowerShell 5.1's 'irm' returns the
   response body as a string with the BOM surviving as a leading
   \ufeff character; 'iex' then evaluates that string and the parser
   chokes on the invisible character before param(), surfacing as a
   cascade of 'The assignment expression is not valid' errors at
   every param default value.

   File body is verified pure ASCII (no character above byte 127),
   so PS 5.1 with no BOM falls back to Windows-1252 decoding which
   is identical to ASCII for our content. Both install paths work:
     - 'irm ... | iex' (canonical one-liner)
     - 'powershell -File install.ps1' (programmatic / desktop bootstrap)

2. New -Commit and -Tag string params for reproducible pinning.

   Higher-precedence variants of -Branch. When set, the repository
   stage clones $Branch (fast partial fetch) and then 'git checkout's
   the exact ref. Precedence: Commit > Tag > Branch. Honoured by all
   three code paths:
     - Update path (existing valid checkout): fetch + checkout
       --detach <commit|tag> instead of checkout + pull.
     - Fresh clone: clone --branch $Branch, then post-clone
       'git checkout --detach' to the requested ref.
     - ZIP fallback: pick archive URL for the most-specific ref
       (commit -> archive/<sha>.zip, tag -> archive/refs/tags/
       <tag>.zip, else archive/refs/heads/<branch>.zip).

   Used by the Hermes desktop's first-launch bootstrap to pin the
   .exe to the exact commit it was built against, so the cloned
   Hermes Agent tree always matches what the .exe was tested with.
   Also enables release-bundle pinning (e.g. Microsoft Store builds
   pinning to a release tag) and CI reproducibility.

3. EAP=Continue wrap around the new pin-step git invocations.

   'git fetch origin <commit>' writes the routine 'From <url>' info
   line to stderr. Under the script's global $ErrorActionPreference
   = 'Stop' that stderr line is wrapped as an ErrorRecord and
   terminates the script even though fetch+checkout actually succeed.
   Same EAP=Stop + native-stderr footgun we hit during the install.ps1
   hardening pass in Install-Uv, Test-Python, _Run-NpmInstall.

   Wrap both the update-path fetch/checkout block AND the post-clone
   pin block in $ErrorActionPreference = 'Continue' (restored in
   finally). Real failures still caught by $LASTEXITCODE checks.

Three issues flagged by the Copilot review on this PR:

1. Double JSON emit on stage failure (Copilot #1, #2). When -Stage <name>
   ran a worker that threw, Invoke-Stage's finally emitted a JSON result
   frame AND the entry-point catch emitted a second error frame --
   producing two concatenated JSON objects on stdout and breaking the
   one-line-per-invocation contract that drivers parse against. Same
   issue applied to -Json mode on a full install (every stage's finally
   plus a final error frame missing duration_ms/skipped).

   Fix: Invoke-Stage's finally now sets $script:_StageEmittedErrorFrame
   when it emits a failure frame; the entry-point catch checks the flag
   and skips its own emit, still exit 1.

2. $prevEAP uninitialized on early try-block throw (Copilot #3). In
   Install-Uv, Test-Python, Test-Node's winget fallback,
   _Run-NpmInstall, and the playwright block, '$prevEAP =
   $ErrorActionPreference' lived as the first statement INSIDE the
   try. If anything between 'try {' and that line threw (Write-Info on
   an unusual host, the npx-finding loop, etc.), the catch's
   'if ($prevEAP) { ... }' restore was a no-op and EAP could remain
   relaxed.

   Fix: hoist '$prevEAP = $ErrorActionPreference' to the line
   immediately before 'try {' in all five sites. Catch's restore is
   now always meaningful regardless of where in the try the throw
   originated.

No change to Invoke-Stage's success path or to the four lint-clean EAP
sites (Test-Node was the only winget-related catch). All 19 metadata
smoke tests still pass.

… ops

Three install.ps1 improvements pulled from the thin-installer work on
bb/gui (PR NousResearch#27822) that benefit the canonical CLI install flow on main:

1. Strip UTF-8 BOM from scripts/install.ps1.

   The canonical 'irm <raw URL> | iex' install flow has been broken
   since commit 4279da4 re-introduced a UTF-8 BOM that PR NousResearch#27224
   had explicitly stripped. PowerShell 5.1's 'irm' returns the
   response body as a string with the BOM surviving as a leading
   \ufeff character; 'iex' then evaluates that string and the parser
   chokes on the invisible character before param(), surfacing as a
   cascade of 'The assignment expression is not valid' errors at
   every param default value.

   File body is verified pure ASCII (no character above byte 127),
   so PS 5.1 with no BOM falls back to Windows-1252 decoding which
   is identical to ASCII for our content. Both install paths work:
     - 'irm ... | iex' (canonical one-liner)
     - 'powershell -File install.ps1' (programmatic / desktop bootstrap)

2. New -Commit and -Tag string params for reproducible pinning.

   Higher-precedence variants of -Branch. When set, the repository
   stage clones $Branch (fast partial fetch) and then 'git checkout's
   the exact ref. Precedence: Commit > Tag > Branch. Honoured by all
   three code paths:
     - Update path (existing valid checkout): fetch + checkout
       --detach <commit|tag> instead of checkout + pull.
     - Fresh clone: clone --branch $Branch, then post-clone
       'git checkout --detach' to the requested ref.
     - ZIP fallback: pick archive URL for the most-specific ref
       (commit -> archive/<sha>.zip, tag -> archive/refs/tags/
       <tag>.zip, else archive/refs/heads/<branch>.zip).

   Used by the Hermes desktop's first-launch bootstrap to pin the
   .exe to the exact commit it was built against, so the cloned
   Hermes Agent tree always matches what the .exe was tested with.
   Also enables release-bundle pinning (e.g. Microsoft Store builds
   pinning to a release tag) and CI reproducibility.

3. EAP=Continue wrap around the new pin-step git invocations.

   'git fetch origin <commit>' writes the routine 'From <url>' info
   line to stderr. Under the script's global $ErrorActionPreference
   = 'Stop' that stderr line is wrapped as an ErrorRecord and
   terminates the script even though fetch+checkout actually succeed.
   Same EAP=Stop + native-stderr footgun we hit during the install.ps1
   hardening pass in Install-Uv, Test-Python, _Run-NpmInstall.

   Wrap both the update-path fetch/checkout block AND the post-clone
   pin block in $ErrorActionPreference = 'Continue' (restored in
   finally). Real failures still caught by $LASTEXITCODE checks.

…nstalls agent-browser

PR NousResearch#65701 fixed a runtime symptom (zombie agent-browser daemon holding a
TCP port after Hermes restart on Windows). While validating NousResearch#65701's fix
on a Windows + NVM + Node 24 environment, a separate sibling install-flow
bug was traced that explains why the Windows desktop population commonly
lands on [email protected] (the version with the zombie bug) instead of
agent-browser@^0.26.0 (where the daemon idle-timeout was wired in commit
284e084). This PR fixes the install-flow bug -- the sibling, not a
duplicate of NousResearch#65701.

Symptom / evidence trail (reproduced on a Windows + NVM + Node 24 env):

* `browser_navigate` returned "Daemon failed to start (port: 127.0.0.1:55478)"
  with a live zombie `daemon.js` (agent-browser v0.17.1 from NVM) on PATH.
* The Hermes process's PATH includes `$HERMES_HOME\node\bin` (the bundled
  prefix `Install-AgentBrowser` populates via `npm install -g --prefix
  $HERMES_HOME\node "agent-browser@^0.26.0"`), but that directory does NOT
  exist on disk for desktop-driven installs.
* So bare-PATH wins and resolves to NVM's stale v0.17.1.

Root cause:

* The desktop Update button in apps/desktop/electron/bootstrap-runner.ts
  (L779) drives install.ps1 per-stage via `install.ps1 -Stage <name>
  -NonInteractive -Json`, iterating stages fetched from `install.ps1
  -Manifest`.
* install.ps1's auto-driven stage list `$InstallStages` (L3501-L3527) had
  13 entries: uv, python, git, node, system-packages, repository, venv,
  dependencies, node-deps, [desktop if -IncludeDesktop], path,
  config-templates, platform-sdks, bootstrap-marker, configure, gateway.
  There was NO `browser` / `agent-browser` stage.
* `Install-AgentBrowser` (L355, the only function that runs `npm install -g
  --prefix`) was reachable only via `install.ps1 -PostInstall` or
  `install.ps1 -Ensure browser` -- neither path is in the auto-driven
  stage sequence.
* Result: a Windows user installing or updating Hermes purely via the
  desktop GUI gets `agent-browser` from whatever global PATH provides
  (stale NVM 0.17.1), not from the ^0.26.0 the install.ps1 code intends.

The Linux/macOS equivalent -- scripts/install.sh's ensure_browser() -- IS
in the install flow via ensure_mode (L2550-2600). The Windows gap is
asymmetric (introduced in NousResearch#27224 when the stage-manifest API was added;
install.sh's ensure_browser predates that).

The fix:

* Add a `browser` stage to $InstallStages, placed AFTER node (the worker
  calls Resolve-NpmCmd which throws if npm is missing) and AFTER desktop
  (if -IncludeDesktop, so a freshly-built Hermes.exe picks up the freshly-
  installed agent-browser on its first relaunch instead of inheriting a
  stale bare-PATH binary). Orderings documented in-line at install.ps1
  L3518-L3531.
* Add a thin `Stage-Browser` worker (L3577-L3609) that delegates to the
  existing `Install-AgentBrowser` function -- extend, don't duplicate (per
  AGENTS.md). The worker:
    - Soft-skips when Node is unavailable (mirrors Stage-Node at L3548-3558):
      browser tools are optional, the install flow MUST NOT abort. The
      skipped distinction is surfaced to the JSON frame via the
      $script:_StageSkippedReason channel Invoke-Stage resets each stage
      (L3609-L3616).
    - Forwards -SkipChromium to Install-AgentBrowser so any future top-level
      param and the existing Find-SystemBrowser internal short-circuit both
      line up.

Idempotency / the stage contract (L3486-L3492):

* npm install -g --prefix is idempotent against a satisfied version range,
  so a second invocation with agent-browser already at ^0.26.0 is a near-
  no-op. Cold installs measured ~5-15s on Windows; warm reinstalls are
  much faster because npm's local cache satisfies the version. The
  protocol version is NOT bumped -- stages are additive (drivers iterate
  the manifest dynamically, per L3492).

Existing-manual-install note (for reviewer consideration):

A user with a manually-installed agent-browser 0.17.1 in their NVM/Node
global continues to have that v0.17.1 in their NVM. After this PR lands,
Hermes-spawned subprocesses get 0.26.0 from the bundled prefix (which is
prepended to PATH). The user's OLD 0.17.1 remains installed in NVM but is
no longer spawned by Hermes -- the bundled-prefix prepend order is what
wins. This is the same ordering install.sh relies on for POSIX.

Tests:

* Python source-level: tests/test_install_ps1_browser_stage.py (11 new
  tests) pins the stage definition at the source-text level so Linux CI
  can verify the structural contract without running PowerShell. Covers:
  stage is declared with the right shape/NeedsUserInput=false/Category;
  Stage-Browser worker delegates to Install-AgentBrowser (not duplicate);
  soft-skip when Node is missing; -SkipChromium flag forwarding; ordering
  after node/node-deps and before configure; ASCII invariant; the worker
  does NOT introduce a process-kill pathway (defense against the same
  Blocking scoping concern tonydwb raised on NousResearch#65701).
* Pester smoke: scripts/tests/test-install-ps1-stage-protocol.ps1 asserts
  'browser' appears in `install.ps1 -Manifest` output, after 'node', and
  before 'configure', with category='install' / needs_user_input=false.

Verified on Windows (PowerShell 7 + Python 3.11.15): all 5 new Pester
assertions PASS, all pre-existing Pester smoke assertions PASS (29 OK /
0 FAIL); 19/19 source-level install.ps1 Python tests pass (1 ascii +
3 node-path + 4 native-stderr + 11 new browser-stage). Linux/macOS not
directly verified; source-level Python tests use only stdlib regex +
Path.read_bytes() so no host-specific behavior is expected, but is not
asserted here. Pester suite is Windows-only by nature (shells out to
powershell.exe).

Coordination with NousResearch#58687 (fix(update): honor configured bootstrap state):
that PR adds the POSIX-side `disabled_toolsets`-aware browser-install
skip. The Windows `Stage-Browser` worker here does NOT currently consult
agent.disabled_toolsets / platform_toolsets -- a Windows desktop user who
ran `hermes tools disable browser` will still get the bundled 0.26.0
install on the next Update, which is exactly the unconditional-install
behavior NousResearch#58687 was filed to stop on Linux. A symmetric Stage-Browser
guard reading $HERMES_HOME\config.yaml's agent.disabled_toolsets field is
the natural follow-up; not bundled here because NousResearch#58687 is itself still
open and that PR is the consumer of the config-awareness infrastructure.

Refs: NousResearch#65701 (sibling), NousResearch#58687 (related -- disabled_toolsets config-
awareness), NousResearch#27224 (manifest API origin), 284e084 (upstream
idle-timeout that 0.26.0 wires).

Three issues flagged by the Copilot review on this PR:

1. Double JSON emit on stage failure (Copilot NousResearch#1, NousResearch#2). When -Stage <name>
   ran a worker that threw, Invoke-Stage's finally emitted a JSON result
   frame AND the entry-point catch emitted a second error frame --
   producing two concatenated JSON objects on stdout and breaking the
   one-line-per-invocation contract that drivers parse against. Same
   issue applied to -Json mode on a full install (every stage's finally
   plus a final error frame missing duration_ms/skipped).

   Fix: Invoke-Stage's finally now sets $script:_StageEmittedErrorFrame
   when it emits a failure frame; the entry-point catch checks the flag
   and skips its own emit, still exit 1.

2. $prevEAP uninitialized on early try-block throw (Copilot NousResearch#3). In
   Install-Uv, Test-Python, Test-Node's winget fallback,
   _Run-NpmInstall, and the playwright block, '$prevEAP =
   $ErrorActionPreference' lived as the first statement INSIDE the
   try. If anything between 'try {' and that line threw (Write-Info on
   an unusual host, the npx-finding loop, etc.), the catch's
   'if ($prevEAP) { ... }' restore was a no-op and EAP could remain
   relaxed.

   Fix: hoist '$prevEAP = $ErrorActionPreference' to the line
   immediately before 'try {' in all five sites. Catch's restore is
   now always meaningful regardless of where in the try the throw
   originated.

No change to Invoke-Stage's success path or to the four lint-clean EAP
sites (Test-Node was the only winget-related catch). All 19 metadata
smoke tests still pass.

… ops

Three install.ps1 improvements pulled from the thin-installer work on
bb/gui (PR NousResearch#27822) that benefit the canonical CLI install flow on main:

1. Strip UTF-8 BOM from scripts/install.ps1.

   The canonical 'irm <raw URL> | iex' install flow has been broken
   since commit 4279da4 re-introduced a UTF-8 BOM that PR NousResearch#27224
   had explicitly stripped. PowerShell 5.1's 'irm' returns the
   response body as a string with the BOM surviving as a leading
   \ufeff character; 'iex' then evaluates that string and the parser
   chokes on the invisible character before param(), surfacing as a
   cascade of 'The assignment expression is not valid' errors at
   every param default value.

   File body is verified pure ASCII (no character above byte 127),
   so PS 5.1 with no BOM falls back to Windows-1252 decoding which
   is identical to ASCII for our content. Both install paths work:
     - 'irm ... | iex' (canonical one-liner)
     - 'powershell -File install.ps1' (programmatic / desktop bootstrap)

2. New -Commit and -Tag string params for reproducible pinning.

   Higher-precedence variants of -Branch. When set, the repository
   stage clones $Branch (fast partial fetch) and then 'git checkout's
   the exact ref. Precedence: Commit > Tag > Branch. Honoured by all
   three code paths:
     - Update path (existing valid checkout): fetch + checkout
       --detach <commit|tag> instead of checkout + pull.
     - Fresh clone: clone --branch $Branch, then post-clone
       'git checkout --detach' to the requested ref.
     - ZIP fallback: pick archive URL for the most-specific ref
       (commit -> archive/<sha>.zip, tag -> archive/refs/tags/
       <tag>.zip, else archive/refs/heads/<branch>.zip).

   Used by the Hermes desktop's first-launch bootstrap to pin the
   .exe to the exact commit it was built against, so the cloned
   Hermes Agent tree always matches what the .exe was tested with.
   Also enables release-bundle pinning (e.g. Microsoft Store builds
   pinning to a release tag) and CI reproducibility.

3. EAP=Continue wrap around the new pin-step git invocations.

   'git fetch origin <commit>' writes the routine 'From <url>' info
   line to stderr. Under the script's global $ErrorActionPreference
   = 'Stop' that stderr line is wrapped as an ErrorRecord and
   terminates the script even though fetch+checkout actually succeed.
   Same EAP=Stop + native-stderr footgun we hit during the install.ps1
   hardening pass in Install-Uv, Test-Python, _Run-NpmInstall.

   Wrap both the update-path fetch/checkout block AND the post-clone
   pin block in $ErrorActionPreference = 'Continue' (restored in
   finally). Real failures still caught by $LASTEXITCODE checks.

Three issues flagged by the Copilot review on this PR:

1. Double JSON emit on stage failure (Copilot #1, #2). When -Stage <name>
   ran a worker that threw, Invoke-Stage's finally emitted a JSON result
   frame AND the entry-point catch emitted a second error frame --
   producing two concatenated JSON objects on stdout and breaking the
   one-line-per-invocation contract that drivers parse against. Same
   issue applied to -Json mode on a full install (every stage's finally
   plus a final error frame missing duration_ms/skipped).

   Fix: Invoke-Stage's finally now sets $script:_StageEmittedErrorFrame
   when it emits a failure frame; the entry-point catch checks the flag
   and skips its own emit, still exit 1.

2. $prevEAP uninitialized on early try-block throw (Copilot #3). In
   Install-Uv, Test-Python, Test-Node's winget fallback,
   _Run-NpmInstall, and the playwright block, '$prevEAP =
   $ErrorActionPreference' lived as the first statement INSIDE the
   try. If anything between 'try {' and that line threw (Write-Info on
   an unusual host, the npx-finding loop, etc.), the catch's
   'if ($prevEAP) { ... }' restore was a no-op and EAP could remain
   relaxed.

   Fix: hoist '$prevEAP = $ErrorActionPreference' to the line
   immediately before 'try {' in all five sites. Catch's restore is
   now always meaningful regardless of where in the try the throw
   originated.

No change to Invoke-Stage's success path or to the four lint-clean EAP
sites (Test-Node was the only winget-related catch). All 19 metadata
smoke tests still pass.

… ops

Three install.ps1 improvements pulled from the thin-installer work on
bb/gui (PR NousResearch#27822) that benefit the canonical CLI install flow on main:

1. Strip UTF-8 BOM from scripts/install.ps1.

   The canonical 'irm <raw URL> | iex' install flow has been broken
   since commit 4279da4 re-introduced a UTF-8 BOM that PR NousResearch#27224
   had explicitly stripped. PowerShell 5.1's 'irm' returns the
   response body as a string with the BOM surviving as a leading
   \ufeff character; 'iex' then evaluates that string and the parser
   chokes on the invisible character before param(), surfacing as a
   cascade of 'The assignment expression is not valid' errors at
   every param default value.

   File body is verified pure ASCII (no character above byte 127),
   so PS 5.1 with no BOM falls back to Windows-1252 decoding which
   is identical to ASCII for our content. Both install paths work:
     - 'irm ... | iex' (canonical one-liner)
     - 'powershell -File install.ps1' (programmatic / desktop bootstrap)

2. New -Commit and -Tag string params for reproducible pinning.

   Higher-precedence variants of -Branch. When set, the repository
   stage clones $Branch (fast partial fetch) and then 'git checkout's
   the exact ref. Precedence: Commit > Tag > Branch. Honoured by all
   three code paths:
     - Update path (existing valid checkout): fetch + checkout
       --detach <commit|tag> instead of checkout + pull.
     - Fresh clone: clone --branch $Branch, then post-clone
       'git checkout --detach' to the requested ref.
     - ZIP fallback: pick archive URL for the most-specific ref
       (commit -> archive/<sha>.zip, tag -> archive/refs/tags/
       <tag>.zip, else archive/refs/heads/<branch>.zip).

   Used by the Hermes desktop's first-launch bootstrap to pin the
   .exe to the exact commit it was built against, so the cloned
   Hermes Agent tree always matches what the .exe was tested with.
   Also enables release-bundle pinning (e.g. Microsoft Store builds
   pinning to a release tag) and CI reproducibility.

3. EAP=Continue wrap around the new pin-step git invocations.

   'git fetch origin <commit>' writes the routine 'From <url>' info
   line to stderr. Under the script's global $ErrorActionPreference
   = 'Stop' that stderr line is wrapped as an ErrorRecord and
   terminates the script even though fetch+checkout actually succeed.
   Same EAP=Stop + native-stderr footgun we hit during the install.ps1
   hardening pass in Install-Uv, Test-Python, _Run-NpmInstall.

   Wrap both the update-path fetch/checkout block AND the post-clone
   pin block in $ErrorActionPreference = 'Continue' (restored in
   finally). Real failures still caught by $LASTEXITCODE checks.