Web Development Iteration prompt now spells out the in-container loop
explicitly: for every failing Playwright test, follow the inner-loop
procedure in role.md together with playwright.md; before every
redeploy run `make exec` and `make run` against the live stack and
fully fix and inspect every failure in-container; the next deploy
iteration only starts once every error is resolved and the fix is
empirically verified. Adds explicit prohibitions on commit/push and
on `ask`-mode approvals, and forbids premature termination — the
iteration is finished only when every role is green end-to-end.
Strips the `Use mode /caveman ultra` postscript from every other
prompt: it was an experimental mode marker that no longer applies
and only added noise. The prompts now stop at the standard
`act autonomously …` sentence.
Reformatted to comply with documentation.md: complete sentences
instead of semicolon chains, no em-dash thought breaks, RFC 2119
keywords used consistently.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Replaces 15 per-shape Bash entries that existed as sandbox auto-allow
workarounds with a single Bash(*) wildcard. Gating responsibility moves
entirely to ask (mutating gh/docker/git verbs) and deny (structurally
destructive shapes). Documentation updated across settings.md,
security.md, and sandbox.md to reflect the new model; em-dash clause
separators removed throughout per documentation.md style rules.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Replaces narrow `Bash(PYTHONPATH=* python *)` with `Bash(*=* python* *)`
so test invocations carrying multiple env-var assignments
(e.g. `PYTHON=python3 PYTHONPATH=. python3 -m unittest ...`) and
`python3` binaries no longer trip the auto-allow heuristic. Sandbox
remains the primary containment layer, so widening the allow shape
adds no new blast radius.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Split docs/agents/action/iteration.md into role.md, playwright.md,
workflow.md, and a scope-focused README. Move Certificate Authority
and Inspect rules into role.md. Rewrite playwright.md as a machine-
oriented state machine with explicit preconditions, procedure, exit
gates, and escape path.
Extract Playwright spec content MUSTs into the new contrib SPOT
docs/contributing/artefact/files/role/playwright.specs.js.md. Reduce
the agent-side spec and env.j2 pages to procedure only. Align the
framework SPOT and the test-e2e-playwright role README with the
central templates/package.json.j2 + files/playwright.config.js model
and images.playwright.version as single source of truth.
Sweep em dashes and sentence-case role headings across all touched
docs per documentation.md and readme_md.md rules.
Three related settings.json changes:
1. Replace narrow `Bash(*=* make *)` with three broader rules covering
any prefix + make, plus composites with `> /tmp/*` redirect and
`2>&1` stderr-merge. Needed so multi-env invocations with quoted
multi-word values (e.g. `FOO=a BAR=b APPS="x y z" make target > /tmp/log 2>&1`)
auto-allow without prompting — the prior single-env rule and the
generic `* > /tmp/*` rules did not compose for this shape.
2. Replace inert `WebFetch(domain:*)` with explicit per-domain entries
for `github.com` and `raw.githubusercontent.com`. Tested empirically:
the wildcard form does not suppress the permission prompt, so a
per-domain allowlist is the only mechanism that works.
3. Add `Bash(git -C *)` to deny. Blocks cross-repo escapes via the
`-C <path>` flag, which otherwise lets the agent reach into other
working trees under `additionalDirectories`. Operators needing a
second repo must open a dedicated session there.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Add `Bash(*=* make *)` so `FOO=bar make <target>` invocations (notably
`make act-workflow` with ACT_WORKFLOW/ACT_MATRIX overrides) auto-allow
without prompting — mirrors the existing `PYTHONPATH=* python *` entry.
Add `*.nektosact.com` to sandbox network allowedDomains so `act` (the
local GitHub Actions runner invoked by `make act-workflow`) can reach
its update/catalog endpoint.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Add ~/.cache/gh to sandbox allowWrite so gh CLI no longer needs an
XDG_CACHE_HOME=/tmp/... prefix, and replace the five XDG-prefixed allow
entries with two generic `* > /tmp/*` redirect entries that cover any
command capturing output into /tmp. Sync sandbox.md / settings.md /
security.md and normalize the allowWrite path enumerations.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Replace the one-off literal allow entry for a specific `gh run download`
invocation with the `XDG_CACHE_HOME=* gh run download *` wildcard that
mirrors the existing `view`/`list` env-prefix precedents, and move the
`Skill(update-config*)` wildcard from `allow` to `ask` so settings-file
mutations (permission rules, hooks, sandbox scope) always prompt — the
skill can otherwise self-escalate the agent's own permission surface.
Document both entries in settings.md with matching Env-Prefix and
Privilege-Escalation rationales.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Removes ~/.ssh from sandbox.filesystem.denyRead so SSH auth (keys, known_hosts, config) succeeds inside the sandbox. Bash(git push*) stays in ask, which becomes the effective control point for pushes.
Documents the trade-off:
- security.md: updates Assumption 9 and adds Assumption 10 covering why ~/.ssh is readable, the cost, and the mitigations (passphrase-protected keys, key rotation on incident, settings.local.json opt-out, SSH-agent-socket alternative).
- sandbox.md: removes ~/.ssh from the Read Restrictions table and links back to Assumption 10.
Collapse six literal XDG_CACHE_HOME=/tmp/gh-cache gh run/issue list
allow entries into XDG_CACHE_HOME=* gh run list *, XDG_CACHE_HOME=*
gh issue list *, gh run *, and gh issue *. Plain gh run * / gh issue *
are safe because every mutating verb is caught by gh * <verb>* ask or
gh issue transfer* deny. Env-prefix variants stay narrow (no broader
than list) because env prefix bypasses subcommand ask/deny patterns.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Allow read-only gh api GETs (plus XDG_CACHE_HOME=* prefix variant) so
log-fetch invocations with env prefix + redirect stop prompting. Route
every implicit-POST shape through ask: -f/-F/--field/--raw-field/--input
body flags and graphql*. Path-based deny rules keep precedence.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Centralise the full GitHub Actions workflow catalog in
docs/contributing/tools/actions.md as the single source of truth, link
pipeline.md and workflow.md to it, and introduce
docs/agents/action/push-trigger-pull.md as the authoritative procedure
for the push to draft CI to ready-for-review loop. Retire the stale
flat docs/contributing/tools/agents/claude.md and drop docs README
page-inventories that violate the readme_md.md scope rule.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Mirrors the XDG_CACHE_HOME=* gh run view * precedent: leading env-prefixes fall out of the auto-allow heuristic. Narrow namespace, sandbox boundaries still bound every write / read / network call.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Parallel to head *: tail is the right-hand-side twin that lives after && / | and would otherwise decline auto-allow.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
- Allow Bash(docker *) under test-only host assumption; keep run/build/push/login in ask
- Catch synonyms: docker container run, docker image build, docker buildx build, docker image push
- New security.md enumerates 9 trust assumptions as single source for rationale
- Shorten settings.md rationales to link by assumption number (DRY)
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
- Add Bash(echo *) and Bash(head *) to allow list for chained stdout/read primitives.
- Move Bash(for *)/Bash(while *)/Bash(until *) to deny: hard-enforces the
CLAUDE.md loop-forbidden rule so violations fail fast and force refactor
to single-command / grep / xargs forms.
- Drop redundant /tmp/* scoping from Bash(XDG_CACHE_HOME=* gh run view *)
and Bash(tee *); sandbox.filesystem.allowWrite is the authoritative
write boundary, so path restrictions on the allow rule added only
maintenance, no safety.
- Update docs/contributing/tools/agents/claude/settings.md to match.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Why: `make` invocations and `… 2>&1 | tee /tmp/<log>` pipelines regularly
fall outside the sandbox auto-allow heuristic and trigger unnecessary
approval prompts. Adding targeted allow entries removes the friction
while keeping the write scope of `tee` bounded to /tmp.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Add explicit Bash allow entries for command shapes that slip past the
sandbox auto-allow heuristic: `gh run list *` (jq pipes in --jq string
confuse the parser), `grep *` (to back the new CLAUDE.md rule against
for-loops around per-file greps), and `wc *` (read-only, any flag).
Reorder permissions.allow strictly ASC so the Bash entries live
alongside the other allow rules in sort order, matching how ask/deny
are laid out. Document the three new entries in settings.md and add
the "prefer flat commands" guidance to CLAUDE.md Code Execution.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Add explicit Bash allow entries for command shapes where sandbox
auto-allow does not fire reliably (env-variable prefix, output
redirect, &&-chained subcommands). Document the rationale in the
settings reference so future additions follow the same pattern.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Copy-paste prompt templates with selection matrix routing each workflow
to its SPOT procedure file. Covers development, iteration, testing,
refactor, debug (pipeline/local/log), commit, PR, and requirements.
Consolidate gh ask/deny entries via namespace-wildcard patterns, cover
compound subcommands (item-*, field-*, mark-template*), and add a lint
test that enforces ASCII-sorted permission and sandbox arrays in
.claude/settings.json.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Sandbox network gate already bounds curl to allowedDomains, so the
interactive prompt added no real protection on top. Remove the rule
and the matching docs row.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Collapse explicit per-host entries into *.parent.tld patterns where the
sandbox only needs reach to a single domain family (Ansible, AWS, Docker,
Gitea, GitHub + githubusercontent, gstatic, hCaptcha, infinito, npm,
pythonhosted, Taiga). Apex domains stay listed bare. Sorted ASC with the
wildcard block first.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Network gate does not honor wildcards, so allowedDomains is enumerated
explicitly for every host the agent + tooling actually need. Adds
~/.cache/pip, ~/.venvs, ~/.ansible to allowWrite so the pre-commit
make test hook can install the editable package and run ansible.
Adds host shell / editor / tooling state files to .gitignore (and the
auto-generated .dockerignore mirror) so they never become part of the
working tree. Switches the dependabot coverage test to git ls-files
with an os.walk fallback that derives its skip list from the live
.gitignore, so it stays correct in container test environments
without .git/ while still ignoring untracked host artefacts.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Adds scripts/install/sandbox.sh, a distro-aware installer for the OS-level
dependencies the Claude Code sandbox backend needs (bubblewrap, socat),
and exposes it through a new `make agent-install` target. Documents the
target in docs/contributing/tools/make.md, scripts/install/README.md, and
the Claude sandbox guide so contributors on a fresh host can bring up a
working sandbox without hand-installing distro packages.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Move docs/contributing/tools/agents/claude.md to claude/settings.md
- Split sandbox explanations into claude/sandbox.md, add claude/README.md as index
- Enable OS-level sandbox ("sandbox.enabled: true") for project-wide filesystem isolation
- Fail-closed on missing backend ("sandbox.failIfUnavailable: true") to prevent false-security state when bwrap/sandbox-exec is absent
- Add deny rules: "git clean*", "git branch -D*"
- Remove redundant "Read(~/.cache/pre-commit/**)" allow entry
- Update CLAUDE.md to instruct agent to summarize live permission state from settings.json at session start
- Update AGENTS.md and common.md references to new doc paths
- Normalize .md formatting per documentation.md rules (em dash removal, headline emojis)
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Split Bash(gh workflow *) into list/view (allow), run (ask), enable/disable (deny)
- Add Bash(break) to allow list
- Remove duplicate /tmp grant from permissions.additionalDirectories; rely on sandbox.filesystem.allowWrite
- Add AGENTS.md "Temporary Files" rule requiring agents to write transient files only under /tmp
- Reorder .claude/settings.json allow entries into logical sections (Git, GitHub CLI, Build, Python, Ansible, Docker, Shell, Read, Web)
- Document new gh workflow entries, break, and Temporary Files cross-link in docs/contributing/tools/agents/claude.md
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Adds `Bash(scripts/*)` so repository helper scripts can be invoked
directly by path (shebang-based), required by inner-loop workflows
such as the Playwright spec rerunner. Adds `Bash(chmod +x scripts/*)`
so newly added scripts can be made executable without approval prompts.
Documents both entries in docs/contributing/tools/agents/claude.md
next to the existing `bash scripts/*` and `source scripts/*` rows.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Adds `Bash(pkill -f *)` to the allowlist so stuck background commands
(pre-commit hooks, `make test`, `git commit`) can be aborted without an
approval prompt. Broad wildcard consciously accepted — narrow per-pattern
entries required repeated allowlist edits for every new abort scenario.
Adds `Read(~/.cache/pre-commit/**)` so pre-commit's stashed patch files
can be reapplied via `git apply` after a manually killed hook run, and
uses `~` instead of a hardcoded username path.
Documents both entries in docs/contributing/tools/agents/claude.md,
including the explicit risk note that `pkill -f` with a regex can
terminate every user-owned process and should use the narrowest pattern
possible.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Replaces two over-specific `Bash(awk ...)` allowlist entries (full paths
and exact patterns that would never match again) with a single
`Bash(awk*)` entry next to `jq*`, consistent with how `grep*`, `find*`,
and `sort*` are allowed. Removes the broken `Read(//var/lib/infinito/logs/**)`
entry: `/var/lib/infinito/` is a named Docker volume defined in
compose.yml and never exists on the host, so host-scoped read access
is meaningless (and the path had a double slash).
Documents the new `awk*` entry in docs/contributing/tools/agents/claude.md
with an explicit note that awk's `system()` and `"cmd" | getline` put
it in the same risk class as the already-allowed `find -exec`.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Adds `Bash(bash scripts/*)` and `Bash(source scripts/*)` to the Claude
Code allowlist so repository helper scripts (e.g. env exports in
scripts/meta/env) can run without approval prompts, while still keeping
arbitrary paths scoped to review.
Documents both permissions in docs/contributing/tools/agents/claude.md,
including the caveat that `source` bypasses the allowlist for nested
commands.
Extends CLAUDE.md with an explicit rule that agents MUST prefer `make`
targets over direct docker/ansible/python/script invocations whenever
an equivalent target exists.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Adds the shell built-in `wait` to the Claude Code allow-list so the agent
can block on background commands started via run_in_background without
prompting for operator approval. Documents the new entry in the Claude
Code settings reference.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Extend .claude/settings.json with git mv, rm, cherry-pick, apply, gh run, gh api, bash -n
- Add make chmod-scripts target and document it in make.md
- Move docs/contributing/tools/agents.md to agents/common.md
- Add docs/contributing/tools/agents/README.md and claude.md with full permission reference in table format
- Fix relative links in common.md after directory move
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
