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 skills add leifericf/agentic-sdk --skill run-review-roundgit clone --depth 1 https://github.com/leifericf/agentic-sdkWrote 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/leifericf/agentic-sdk/run-review-round)<a href="https://agentmods.dev/skills/leifericf/agentic-sdk/run-review-round"><img src="https://agentmods.dev/badge/skills/leifericf/agentic-sdk/run-review-round.svg" alt="Measured on agentmods" 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 | $0.00000 | $0.01949 |
| Opus 5 | $0.00000 | $0.00975 |
| Sonnet 5 | $0.00000 | $0.00390 |
| Haiku 4.5 | $0.00000 | $0.00195 |
Grade A, and why
run-review-round 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 8d 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 — 150 lines — stays where its author put it; the contents beside it link to each section on GitHub.
run-review-round
One round: everything from a stable stack to that stack plus this
round's findings fixed and verified. Run by the review-round-runner
agent.
The round's clerical work is owned by the deterministic spine; the
skill calls the spine tasks by name, which answer at the project's
spine-presence level (full, thin, or none, per references/spine.md).
At full spine they fold the working dir deterministically; at thin or
none the round-runner folds the same data by return value, keeping the
discipline intact. Sub-agent dispatch follows the runtime adaptation
rule in references/orchestration.md: fan out where the Task tool
exists, else load the named recipe (check-*, apply-findings,
verify-lanes) inline.
Inputs from the dispatch: the stack (referenced by its head's change id, a stack head not a branch), the scope (module shards), the round number, and the run slug if the run keeps a resume checkpoint.
Working state
The round carries its working state in the spine working dir
(.<project>/, per references/spine.md) at full spine, and in your
context at thin or none: the reviewer findings written to the
findings/ pool; the punch list the triage task produces; the editor
returns (LANDED, needs-cross-module, FAILED) and changelog lines
you collect; the escalations list, started empty and grown by a double
FAILED on one finding; for audit-code, the out-of-scope items set
aside. A long run may keep a minimal gitignored resume checkpoint under
~/.agentic-sdk/<project>/runs/<slug>/ recording only what is done; never
committed, never the hand-off medium.
Batching
Fan out in module-batches, not one giant wave. A batch is one module's worth of dispatches, run in parallel; wait for the batch, collect the returns, then dispatch the next. Reviewers: one batch per shard, one reviewer per applicable dimension (typically 3 to 7 concurrent). Editors: one batch per level and wave, one editor per module with findings at that level (typically 1 to 5 concurrent). Keep concurrency modest, around 5 to 7 per batch; split a larger batch sequentially. This bounds both concurrency and orchestrator context (N compact returns per batch, not 50 at once).
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.
- 8d ago First seen · 150 lines · 0 tokens per session scan A 19120c7dacae
run-review-round is a skill published in the GitHub repository leifericf/agentic-sdk (5 stars, last pushed 6d ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 1,949 tokens. 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
autoreview
Pre-commit/ship code review: Codex default; optional Claude or Pi.
rework-rate
Measure and interpret PR rework rate — the emerging 5th DORA metric.
omh-code-review
This is a Hermes-native code-review workflow skill.
revdiff-plan
Review the last Codex assistant message (plan, analysis, or proposal) with inline annotations in a TUI overlay. Extracts the most recent response from Codex rollout files and opens it in revdiff for review and annotation. Activates on "revdiff-plan", "review plan with revdiff", "annotate plan", "review last response"…
code-reviewer
Code review specialist focused on patterns, bugs, security, and performance.
agent-teams-simplify-and-harden
Implementation + audit loop using parallel agent teams with structured simplify, harden, and document passes. Spawns implementation agents to do the work, then audit agents to find complexity, security gaps, and spec deviations, then loops until code compiles cleanly, all tests pass, and auditors find zero issues or…