Skip to content

docs: start action safety phase ledger - #227

Merged
delete-cloud merged 1 commit into
mainfrom
codex/action-safety-g25-current-state
May 19, 2026
Merged

delete-cloud merged 1 commit into
mainfrom
codex/action-safety-g25-current-state

Conversation

@delete-cloud

@delete-cloud delete-cloud commented May 19, 2026 •

Copy link
Copy Markdown
Owner

Summary

  • add the G25-G37 action safety phase ledger
  • document current local/cloud file, patch, shell, approval, and workspace archive surfaces
  • add the G25 task packet for bounded implementation/review/verification

Verification

  • test -f docs/action_safety/GOAL_PROGRESS.md
  • test -f docs/action_safety/CURRENT_STATE.md
  • test -f .opencode/prompts/tasks/action-safety-g25-current-state.md
  • uv run pytest tests/coding_agent/tools/test_file_ops.py tests/coding_agent/tools/test_shell.py tests/approval/test_policy.py tests/coding_agent/environment/test_workspace_archive.py -v
  • uv run pytest tests/coding_agent/test_context_system_smoke.py -v
  • uv run pytest tests/agentkit/runtime/test_pipeline.py -k "build_context or runtime_stage_spans" -v

Local Review

  • local subagent review found one P2 docs-correctness issue around local vs cloud shell behavior
  • fixed by splitting local/cloud tool behavior in CURRENT_STATE and task packet
  • reran target verification after the fix

Summary by CodeRabbit

  • Documentation
    • Added comprehensive documentation of current action safety and workspace execution system behavior and safety boundaries.
    • Established development phase ledger tracking planned improvements across multiple release cycles.

Review Change Stack

@coderabbitai

coderabbitai Bot commented May 19, 2026 •

Copy link
Copy Markdown
Contributor

Caution

Review failed

Pull request was closed or merged during review

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info
⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro Plus

Run ID: f07bd25f-1514-41e8-bfe9-a64e593ff70b

📥 Commits

Reviewing files that changed from the base of the PR and between ad68b80 and b64b811.

📒 Files selected for processing (3)
  • coding-agent/.opencode/prompts/tasks/action-safety-g25-current-state.md
  • coding-agent/docs/action_safety/CURRENT_STATE.md
  • coding-agent/docs/action_safety/GOAL_PROGRESS.md

📝 Walkthrough

Walkthrough

This PR establishes the G25 baseline audit for action-safety and workspace execution by creating three documentation artifacts: a task specification defining scope and loop conditions, a comprehensive CURRENT_STATE baseline documenting system boundaries, tool surfaces, approval policy, and workspace archive behavior, and a phase ledger tracking goals G25–G37 with verification rules and gaps identified for future work.

Changes

Action Safety Phase Ledger: G25 Current State Baseline

Layer / File(s) Summary
Audit Task Specification
coding-agent/.opencode/prompts/tasks/action-safety-g25-current-state.md
G25 audit scope, relevant ADR/code references, postmortem routing rules, target tests/commands, and loop/stop conditions governing implement-review-fix cycles with minimal doc changes and max one cycle policy.
System Boundaries and File Tooling
coding-agent/docs/action_safety/CURRENT_STATE.md (header, ownership, file tools)
Ownership boundaries between agentkit (generic tool/runtime) and coding_agent (concrete policy/execution); local file tools (file_read, file_write, file_replace, glob_files, grep_search) with workspace-root enforcement; cloud file tool delegation; file_patch unified-diff behavior with hunk parsing and temp-file replace, lacking pre-application plan and safety control contracts.
Shell Execution and Session State Wiring
coding-agent/docs/action_safety/CURRENT_STATE.md (bash_run, core tool wiring)
Local bash_run safety (empty-command rejection, metacharacter rejection, cwd/env sync); cloud delegation with less parsing; core tool wiring that injects session state (cd/export) and syncs back across local or cloud environments as the app-layer boundary for future enforcement.
Approval Policy and Workspace Archives
coding-agent/docs/action_safety/CURRENT_STATE.md (approval, archives)
Tool-name based approval policy (YOLO/INTERACTIVE/AUTO; default safe names) lacking file-path, patch-size, destructive-command, or cwd/env-risk classification; workspace archive safety (excluded paths, symlink/traversal defenses, temp extraction, .git reconciliation).
Test Coverage, Gaps, and Implementation Rules
coding-agent/docs/action_safety/CURRENT_STATE.md (coverage, gaps, rules)
Existing deterministic test coverage for file ops, shell, policy, and archive; identified gaps for G26–G37 (ADR, typed edit/patch controls, command policy contracts, observability, local snapshot/restore MVP, action-risk approval, end-to-end smoke); implementation constraints (preserve agentkit genericity, keep policy in coding_agent, maintain JSONL/durable semantics, deterministic tests, observability scope, exclude schedules/Docker sandbox in this phase).
Phase Progress Ledger
coding-agent/docs/action_safety/GOAL_PROGRESS.md
G25–G37 phase ledger with phase constraints, goal map, and detailed G25 entry including status, intended files, verification commands, stop criteria, postmortem routing, tests run, results, and remaining risks.

Estimated code review effort

🎯 1 (Trivial) | ⏱️ ~3 minutes

Possibly related PRs

  • delete-cloud/Coding-Agent-Loop#100: Establishes ADR for cloud workspace execution semantics and CloudEnvironment/CloudWorkspaceClient concepts that form the baseline for this PR's CURRENT_STATE documentation of action-safety ownership boundaries and tool delegation.

Poem

🐰 A ledger takes shape, baseline recorded clear,
G25 audit maps the action-safety frontier,
File tools, shell bounds, approval and archive aligned,
Gaps for G26–G37 documented and signed.
One cycle to verify, then onward we bound! 🚀

🚥 Pre-merge checks | ✅ 5
✅ Passed checks (5 passed)
Check name Status Explanation
Description Check ✅ Passed Check skipped - CodeRabbit’s high-level summary is enabled.
Title check ✅ Passed The title 'docs: start action safety phase ledger' clearly and concisely summarizes the main change—introducing documentation for the action safety phase ledger (G25–G37). It directly reflects the primary objective of the changeset.
Docstring Coverage ✅ Passed No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check ✅ Passed Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check ✅ Passed Check skipped because no linked issues were found for this pull request.

✏️ Tip: You can configure your own custom pre-merge checks in the settings.

✨ Finishing Touches
🧪 Generate unit tests (beta)
  • Create PR with unit tests
  • Commit unit tests in branch codex/action-safety-g25-current-state

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

Comment @coderabbitai help to get the list of available commands and usage tips.

@delete-cloud
delete-cloud merged commit cc4ba94 into main May 19, 2026
1 check was pending
@delete-cloud
delete-cloud deleted the codex/action-safety-g25-current-state branch May 19, 2026 14:19
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant