Getting it into your agent
This one installs as part of its plugin. Adding the marketplace and installing the plugin brings it with everything else the plugin ships.
/plugin marketplace add chan4lk/specclaw/plugin install specclawWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/chan4lk/specclaw/bf-replay)<a href="https://agentmods.dev/skills/chan4lk/specclaw/bf-replay"><img src="https://agentmods.dev/badge/skills/chan4lk/specclaw/bf-replay/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/chan4lk/specclaw/bf-replay"><img src="https://agentmods.dev/badge/skills/chan4lk/specclaw/bf-replay.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What 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 5.1 | $0.00799 | $0.12354 |
| Opus 5 | $0.00400 | $0.06177 |
| Sonnet 5 | $0.00160 | $0.02471 |
| Haiku 4.5 | $0.00080 | $0.01235 |
Grade A, and why
bf-replay 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 5d 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 — 299 lines — stays where its author put it; the contents beside it link to each section on GitHub.
specclaw bf-replay
First, run specclaw-ensure-init .specclaw — idempotently creates .specclaw/ if it doesn't exist (silent if already initialized).
Compares the new app's actual behaviour against .specclaw/baseline/'s captured golden-master fixtures. This is not /specclaw:verify — verify checks a change against its own spec/tasks; replay checks the rebuild against the legacy app's own recorded behaviour. Neither command modifies the other; do not merge this flow into verify.
Invocation:
/specclaw:bf-replay <change-name> # fixtures behind this change's cited BL item; retains evidence
/specclaw:bf-replay --item BL-### # fixtures behind ONE backlog item; no change directory needed
/specclaw:bf-replay --module MOD-### # every fixture tagged with that module; retains evidence
/specclaw:bf-replay --all # full regression: every fixture in the manifest; retains evidence
/specclaw:bf-replay <change-name> --keep # accepted for compatibility — identical to the default
/specclaw:bf-replay <change-name> --discard # opt this one run out of evidence retention (never for an acceptance run)
/specclaw:bf-replay <change-name> --prune-evidence <n> # keep only the newest n evidence runs for this change; no new replay run
Four selection scopes, smallest to largest: --item BL-### → <change-name> → --module MOD-### → --all. --item is the single-item acceptance gate — accept one backlog item's behaviour on its own, before any change exists for it and without inventing one. --module is the module-at-a-time gate: it replays every fixture the manifest tags with that module, so a large legacy system can be migrated and behaviourally signed off one flow at a time instead of all-or-one-change. All four are selection only — verdict logic, the four-step verdict order, PROVISIONAL semantics, ST-### taint mechanics, and every exit code are identical across every one of them.
The four scopes are mutually exclusive. A positional <change-name> combines with no scope flag, and the flags combine with nothing. Retention qualifiers (--keep, --discard, --prune-evidence <n>) qualify a run rather than selecting one, and combine with any scope.
Never decide the scope by reading the invocation yourself — resolve it mechanically first:
target="$(specclaw-bf-replay parse-target <the user's flags and arguments, verbatim>)"
It prints the one canonical target token (<change-name> | BL-### | MOD-### | --all) or fails loudly naming the conflict. If it exits non-zero, surface its stderr verbatim and stop. Pass $target as the target argument to resolve, render, and finalize — every subcommand accepts all four forms in the same argument position, so no separate flag is threaded through five call sites.
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.
- 5d ago First seen · 299 lines · 0 tokens per session scan A 4e82f1256b60
bf-replay is a skill published in the GitHub repository chan4lk/specclaw (12 stars, last pushed 6d ago), licensed MIT. It adds 799 tokens to every session and 12,354 once invoked, about $0.0040 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 2026-09-05.
Other skills, from other repositories
run-preflight
A release pre-check workflow that builds the mini-app, runs automated checks, explains errors, and lists review items that require a person to test the app on a real device. A preflight check is a final inspection before submission.
ai-dev-team
The AI Dev Team contract — how work is run in any project using this framework: proportional process, TDD, behaviour-only tickets, which process skills exist and when to reach for them, and the git conventions. Load this at the START of any development task, when picking an agent or handing off between roles, and…
product-verification
Drive the running product and verify it works end-to-end. Use this skill after a deploy, before a release, or when the user says "verify", "smoke test", "does it work", "test the flow", "check staging", "end-to-end test", or "drive the product". Also use when the user wants to verify a specific user journey (signup…
smoke-test
Post-sprint validation. Verifies the application works as expected after a sprint completes. Use after any sprint or deployment, or when the user says "smoke test", "verify it works", "test the app", "does it still work", "regression check", "post-deploy check", or "sanity check". Invoked by SP-QA during post-sprint…
implement
Use in the Implement phase whenever writing or editing production Java code, or fixing a bug, in a Spring/Spring Boot project. Enforces test-first (red-green-refactor), executes the approved plan step by step, and honors the project's path-scoped tech-stack rules and the task's enforcement set. Preloaded into…
systematic-debugging
4-phase root cause debugging: understand bugs before fixing.