On Windows 10/11 without Python installed from the Microsoft Store, the
"App Execution Alias" stubs at %LOCALAPPDATA%\Microsoft\WindowsApps\python.exe
and python3.exe are symlinks to AppInstallerPythonRedirector.exe. These
stubs neither launch Python nor honor `-c`; calls print a bare "Python "
line and exit, silently breaking every JSON-parsing step in observe.sh.
Net effect: observations.jsonl is never written, CLV2 appears installed
correctly, and the only residual artifact is `.last-purge`.
This commit:
1. Adds `_is_windows_app_installer_stub` helper that detects the stub
via `command -v` output and optional `readlink -f` resolution.
2. Teaches `resolve_python_cmd` to skip stub candidates and fall
through to the next real interpreter (typically C:\...\Python3xx\python.exe).
3. Exports the stub-aware CLV2_PYTHON_CMD before sourcing
detect-project.sh, which already honors an already-set value,
so the shared helper does not re-resolve and re-select the stub.
POSIX-compatible. No behavior change on macOS / Linux / WSL where no
such stub exists.
Refs: observations.jsonl empty on Windows Claude Desktop users.
The merged hero was being clipped at the bottom by the Puppeteer capture
because the HTML body used flex-centering with 24px padding, shifting the
stage below the viewport top.
- Captures now flush to (0,0) via a min-width 1300px media-query wrapper
so the in-browser preview keeps its padding but the capture viewport
does not.
- Shortens bottom-row labels so the stats row no longer overlaps the foot
line at 1200px:
Catalog, Harnesses, Rust plane, MIT / npm: ecc-universal · AgentShield
No other content changes.
Co-authored-by: livlign <livlign@users.noreply.github.com>
* fix: resolve Claude Code Bash hook "cannot execute binary file" on Windows
Root cause in ~/.claude/settings.local.json (user-global):
1. UTF-8 BOM + CRLF line endings left by patch_settings_cl_v2_simple.ps1
2. Double-wrapped command "\"bash.exe\" \"wrapper.sh\"" broke Windows
argument splitting on the space in "Program Files", making bash.exe
try to execute itself as a script.
Fix:
- Rewrite settings.local.json as UTF-8 (no BOM), LF, with the hook command
pointing directly at observe-wrapper.sh and passing "pre"/"post" as a
positional arg so HOOK_PHASE is populated correctly in observe.sh.
Docs:
- docs/fixes/HOOK-FIX-20260421.md — full root-cause analysis.
- docs/fixes/apply-hook-fix.sh — idempotent applier script.
* docs: addendum for HOOK-FIX-20260421 (v2.1.116 argv duplication detail)
- Documents Claude Code v2.1.116 argv duplication bug as the underlying
cause of the bash.exe:bash.exe:cannot execute binary file error
- Records night-session fix variant using explicit `bash <path>` prefix
(matches hooks.json observer pattern, avoids EFTYPE on Node spawn)
- Keeps morning commit 527c18b intact; both variants are now documented
---------
Co-authored-by: suusuu0927 <sugi.go.go.gm@gmail.com>
The SessionStart hook injects the most recent *-session.tmp as
additionalContext labelled only with 'Previous session summary:'.
After a /compact boundary, the model frequently re-executes stale
slash-skill invocations it finds inside that summary, re-running
ARGUMENTS-bearing skills (e.g. /fw-task-new, /fw-raise-pr) with the
last ARGUMENTS they saw.
Observed on claude-opus-4-7 with ECC v1.9.0 on a firmware project:
after compaction resume, the model spontaneously re-enters the prior
skill with stale ARGUMENTS, duplicating GitHub issues, Notion tasks,
and branches for work that is already merged.
ECC cannot fix Claude Code's skill-state replay across compactions,
but it can stop amplifying it. Wrap the injected summary in an
explicit HISTORICAL REFERENCE ONLY preamble with a STALE-BY-DEFAULT
contract and delimit the block with BEGIN/END markers so the model
treats everything inside as frozen reference material.
Tests: update the two hooks.test.js cases that asserted on the old
'Previous session summary' literal to assert on the new guard
preamble, the STALE-BY-DEFAULT contract, and both delimiters. 219/219
tests pass locally.
Tracked at: #1534
* fix(gateguard): rewrite routineBashMsg to use fact-presentation pattern
The imperative 'Quote user's instruction verbatim. Then retry.' phrasing
triggers Claude Code's runtime anti-prompt-injection filter, deadlocking
the first Bash call of every session. The sibling gates (edit, write,
destructive) use multi-point fact-list framing that the runtime accepts.
Align routineBashMsg with that pattern to restore the gate's intended
behavior without changing run(), state schema, or any public API.
Closes#1530
* docs(gateguard): sync SKILL.md routine gate spec with new message format
CodeRabbit flagged that skills/gateguard/SKILL.md still described the
pre-fix imperative message. Update the Routine Bash Gate section to
match the numbered fact-list format used by the new routineBashMsg().
