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/drafthq/draft/assist-reviewnpx skills add drafthq/draft --skill assist-reviewgit clone --depth 1 https://github.com/drafthq/draftWrote 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/drafthq/draft/assist-review)<a href="https://agentmods.dev/skills/drafthq/draft/assist-review"><img src="https://agentmods.dev/badge/skills/drafthq/draft/assist-review.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 | $0.00067 | $0.00781 |
| Opus 5 | $0.00034 | $0.00391 |
| Sonnet 5 | $0.00013 | $0.00156 |
| Haiku 4.5 | $0.00007 | $0.00078 |
Grade A, and why
assist-review 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 4d 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 — 51 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Draft Assist Review: Human-in-the-Loop Gateway
Help human reviewers effectively review an executed track without shifting the entire cognitive burden onto them.
Red Flags - STOP if you're
- Conducting standard unit tests; use
/draft:reviewfor that. - Fixing code rather than explaining logic and risk profiles to the human.
- Reviewing output without first summarizing the source
spec.mdintent.
Workflow Constraints
-
Context Extraction:
- Load the track's
spec.mdandplan.md. - Load the track's
hld.mdandlld.mdif present — extract Key Design Decisions, Alternatives Considered, and (LLD) class invariants and error policies. - Re-summarize the Intent of what this track was supposed to achieve in exactly two sentences.
- Load the track's
-
Blast Radius Isolation:
- Scan the
git diffgenerated by the target track. - Separate trivial edits (naming, routing adjustments, formatting) from Structural Edits (schema updates, concurrency alterations, middleware auth changes, API surface changes).
- Present structural edits first — these are where review time should be spent.
- Scan the
-
Generate the Human Helper Guide:
- Instead of a traditional bug hunt, generate an executive summary containing a Risk Assessment.
- For each structural edit, cite the originating decision: "I chose
[Pattern]for[File/Function]because hld.md §Key Design Decisions / Alternatives Considered selected it over [rejected alternative]. You should specifically scrutinize lines X–Y because they touch [global state / auth boundary / shared schema]." Fall back toarchitecture.mdonly when no HLD exists. - Highlight any code that violates an LLD §Classes and Interfaces invariant (thread safety, idempotency, ordering) or contradicts an HLD §Checklist claim (e.g., "code introduces a new shared mutex but HLD declared single-writer").
- Highlight any code that touches shared state, auth boundaries, data persistence, or concurrency.
-
Knowledge Base Verification:
- Verify if any pattern implemented violates the
draft/guardrails.mdlearned anti-patterns. - Verify the diff is consistent with hld.md/lld.md commitments. If the diff makes structural changes not reflected in HLD §Detailed Design or LLD §Classes/Interfaces: flag for the reviewer that HLD/LLD must be amended (
/draft:change) before merge. - If so, specifically direct the human reviewer to veto the change unless an ADR is created via
/draft:adr.
- Verify if any pattern implemented violates the
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.
- 4d ago First seen · 51 lines · 67 tokens per session scan A 31131927ef10
assist-review is a skill published in the GitHub repository drafthq/draft (40 stars, last pushed 16d ago), licensed MIT. It adds 67 tokens to every session and 781 once invoked, about $0.0003 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 2026-08-30.
Other skills, from other repositories
memorix-memory
Use when prior workspace context, past decisions, solved bugs, handoff state, or durable project knowledge would help a coding task.
memorix
Use when Claude Code needs Memorix shared memory, reasoning, Git Memory, mini-skills, session handoff, orchestration coordination, or integration troubleshooting.
memorix-mini-skills
Use when durable project knowledge, gotchas, workflows, or repeated fixes should become reusable agent guidance instead of ordinary memory.
memorix-orchestrate
Use when a main agent needs Memorix to coordinate explicit subagent work through tasks, handoffs, messages, file locks, or the orchestrate CLI.
memorix-reasoning
Use when a technical decision, trade-off, rejected alternative, architecture rationale, or design risk should be recorded or recovered.
memorix-troubleshooting
Use when Memorix MCP, setup, project binding, HTTP control plane, hooks, skills, or agent integration is missing, stale, or failing.