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.
git clone --depth 1 https://github.com/xiaolai/nlpmWrote 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/commands/xiaolai/nlpm/vocab-drift)<a href="https://agentmods.dev/commands/xiaolai/nlpm/vocab-drift"><img src="https://agentmods.dev/badge/commands/xiaolai/nlpm/vocab-drift.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 | $0.00052 | $0.00793 |
| Opus 5 | $0.00026 | $0.00396 |
| Sonnet 5 | $0.00010 | $0.00159 |
| Haiku 4.5 | $0.00005 | $0.00079 |
Grade A, and why
vocab-drift 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 — 78 lines — stays where its author put it; the contents beside it link to each section on GitHub.
User Input
$ARGUMENTS
Workflow
Step 1: Parse target
| Input | Behavior |
|---|---|
| (empty) | Target = current working directory |
| absolute path | Target = that path |
| relative path | Target = <cwd>/<path> |
If the target does not exist or is not a directory → "Target path not found: {path}". Stop.
Step 2: Discover artifacts
Use commands/shared/discover.md against the target path to collect all NL artifacts. Filter to Category A (plugin) and Category B (project config) — these are where vocabulary lives. Skip Category F (memory files) — they are user-specific and not part of the project's published surface.
If the discovery returns fewer than 5 artifacts → "Target has minimal NL surface ({N} artifacts). Vocabulary drift analysis is uninformative on small corpora; re-run when the corpus grows past 5 artifacts." Stop.
Step 3: Dispatch the scanner
Invoke the nlpm:vocab-drift-scanner agent via the Task tool.
Pass the agent:
- The full list of artifact paths from Step 2
- The contents of every artifact (read with
Readbefore dispatching) - The path to
skills/*/vocabulary/registry.yamlif one exists in the target (useGlobto find it). The scanner readscross_scope_homonyms.verbsfrom this file to suppress false positives on already-declared homonyms.
If the artifact set is large (>30 files), batch into groups of ≤20 artifacts per scanner invocation. The agent's clustering algorithm scales linearly in the number of unique terms, not files, so batches can run in parallel via separate Task dispatches.
Step 4: Merge results (if batched)
If multiple batches ran, merge findings:
- Two findings with overlapping term sets (≥1 shared term in the cluster) → merge into one finding. Sum occurrence counts; combine file lists; take the higher confidence.
- Re-rank the merged set by the criteria in the scanner's Step 5.
- Cap final output at 20 findings.
Step 5: Report
Print the scanner's output verbatim. Then append a single line:
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 · 78 lines · 52 tokens per session scan A c9a763f33751
vocab-drift is a command published in the GitHub repository xiaolai/nlpm (135 stars, last pushed today), licensed ISC. It adds 52 tokens to every session and 793 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 commands, from other repositories
checklist
Generate a custom checklist for the current feature based on user requirements.
clarify
Identify underspecified areas in the current feature spec by asking up to 5 highly targeted clarification questions and encoding answers back into the spec.
specify
Create or update the feature specification from a natural language feature description.
converge
Assess the current codebase against the feature's spec, plan, and tasks, then append any remaining unbuilt work as new tasks to tasks.md so implement can complete it.
implement
Execute the implementation plan by processing and executing all tasks defined in tasks.md.
analyze
Perform a non-destructive cross-artifact consistency and quality analysis across spec.md, plan.md, and tasks.md after task generation.