Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add skills/dmatut7/codex-flow/codex-skill-parallel-fixnpx skills add Dmatut7/codex-flow --skill codex-skill-parallel-fixgit clone --depth 1 https://github.com/Dmatut7/codex-flowWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00162 | $0.00901 |
| Opus 5 | $0.00081 | $0.00451 |
| Sonnet 5 | $0.00032 | $0.00180 |
| Haiku 4.5 | $0.00016 | $0.00090 |
Grade A, and why
parallel-fix scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 40 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Parallel Fix
Speed up the fix/implement half of the loop. A normal agent fixes a list serially (explore → fix → test → commit, repeat) — slow. This skill generates every fix concurrently, then integrates + verifies once.
Engine / backend / sandbox / journal mechanics: reuse the dynamic-workflow skill. This skill carries only the method.
The key idea (and its honest limit)
The slow, parallelizable part is "analyze each finding + write the correct patch" — independent per finding. So fixers run in parallel but READ-ONLY, each returning a unified diff (not editing the tree — parallel writers to one repo race and the engine will reject them). A single serial step then applies the diffs, resolves conflicts, and runs the full suite once.
Honest limits — say these to the user, don't hide them:
- Integration + verify is serial (one suite run on the merged result) — that's the bottleneck you can't parallelize away.
- Overlapping fixes (same file/lines) can't be blindly merged — the apply step puts conflicts in a
conflictedlist for manual handling. - Parallel fixers don't see each other, so a fix may assume code another changed — the single full-suite verify catches breakage; a reconciliation pass may be needed.
- Realistic speedup ≈ 2–4× (more independent findings → more speedup), NOT N×.
Steps
Generate .codex-flow/generated/<slug>.fix.workflow.ts from references/fix-template.workflow.ts (import-free, plain JSON Schema), set FINDINGS to the real list (from the audit / the user), then codex-flow run it. The phases:
- Propose (
parallel,sandbox: "read-only") — one agent per finding. Each reads the code and returns the smallest correct unified diff + ariskrating. Read-only ⇒ safe to share onecwd. Routerisk: "high"items (concurrency / resume / budget / determinism) out of the blind-parallel path — recommend the user fix those with single careful TDD (a reproducing test first), because that's where parallel-blind fixes regress. - Apply + verify (
phase, single agent,sandbox: "workspace-write",cwd: process.cwd()) — apply the diffs in a conflict-minimizing order; record un-mergeable ones underconflicted; run the project's full test + typecheck once; reporttestPassedhonestly.
What ships with it
3 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 2d ago First seen · 40 lines · 162 tokens per session scan A 1784548b89ce
parallel-fix is a skill published in the GitHub repository Dmatut7/codex-flow (9 stars, last pushed 2mo ago), licensed MIT. It adds 162 tokens to every session and 901 once invoked, about $0.0008 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
codex-workflows
Run a dynamic-workflow script on a local Codex App Server — orchestrate many Codex / GPT agents (the agent / parallel / pipeline / phase / budget DSL) instead of Claude subagents, for codebase audits, large migrations, and multi-agent review or research. Give it one or two rough sentences and it compiles the right…
final-release-review
Perform pre-release planning or a final release-candidate review for openai-agents-python by comparing the target with the previous remote tag, determining the minimum compatible release type, auditing regressions and contract changes, reviewing open documentation PR coverage, drafting minor-release Key Changes, and…
playwright
Use when the task requires capturing or automating a real browser from the terminal.
implementation-final-review
Perform the repository's risk-tiered independent final review before implementation completion. Use only when explicitly invoked or when repository instructions require it after behavior-impacting implementation work; audit the complete task diff, supported contracts, lifecycle and security boundaries, complexity, and…
maintainer-review
Assess an openai-agents-python GitHub issue or pull request as a maintainer. Use to verify the claimed need and practical impact, compare supported alternatives or competing approaches, separate code quality from repository readiness, recommend the maintainer action, and draft a copy-ready comment when evidence…
agent-framework-py-release
Use when cutting a Python release for the microsoft/agent-framework monorepo. Triggers on "bump py versions", "cut a python release", "prepare release PR for python", "release py packages", "bump python to X.Y.Z", or similar requests to bump Python package versions and prepare a release PR. Handles all four lifecycle…