contextd-eval

A read-only command that reviews trace files from contextd runs and produces a Markdown report about the effectiveness of a project wiki and its context-selection process.

In plain words
What is it for?
Use it to compare runs, measure wiki coverage, find hallucination and violation hotspots, rank knowledge gaps, and save a dated evaluation report.
Why use it?
It makes it easier to see whether the wiki supplies the right information, where knowledge is missing, and where the system gives unsupported answers or blocks context.

Command for Claude Code

Install

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.

agentmods
npx agentmods add commands/philngt/contextd/contextd-eval
Clone the repo
git clone --depth 1 https://github.com/philngt/contextd

Made for: Claude Code.

Per session 0 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 2,090 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5 $0.00000 $0.02090
Opus 5 $0.00000 $0.01045
Sonnet 5 $0.00000 $0.00418
Haiku 4.5 $0.00000 $0.00209

Measured yesterday against content hash 77866ad9151b, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

contextd-eval 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 yesterday.

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.

.claude/commands/contextd-eval.md · 242 lines

How it starts

The opening of the file, as written. The whole thing — 242 lines — stays where its author put it; the contents beside it link to each section on GitHub.

/contextd-eval — Evaluate Pipeline Effectiveness

Aggregate trace JSON từ nhiều run dưới {project_dir}/.contextd/runs/ → 1 báo cáo Markdown đo hiệu quả wiki: coverage, hallucination rate, top knowledge gaps, context-selector block rate, violation hotspots.

One-shot: 1 lệnh → 1 file Markdown trong {ws}/reports/. Read-only — KHÔNG sửa runs/. Reference: observability.md, run-trace.schema.json.


Input

Arg Required Notes
--out optional Đường dẫn output. Default: {ws}/reports/contextd-eval-{YYYY-MM-DD}.md. Cùng ngày đã có file → suffix -{HHMMSS}.
--since optional Filter runs có ts_start ≥ ngày này (YYYY-MM-DD). Default: tất cả.
--top optional Top-N cho gap/violation/block ranking. Default: 5.

Bước 0 — Workspace check

Theo workspace-resolution.md Profile A. Set: config_dir (= project_dir), workspace_active, effective_knowledge_root, {ws}. Đồng thời:

  • Validate {project_dir}/.contextd/runs/ tồn tại. Nếu không → STOP No runs found. Chạy /use-contextd ít nhất 1 lần trước..

Bước 1 — Discover runs

Glob {project_dir}/.contextd/runs/*/. Mỗi sub-dir = 1 run. Lưu danh sách run_ids.

Với mỗi run:

  1. Đọc run.json (nếu có) — lấy ts_start, user_task, workspace_at_run, stages_completed.
  2. Filter:
    • Workspace lock: bỏ qua run có workspace_at_run != workspace_active (KHÔNG leak cross-workspace metric).
    • Date filter: nếu --since có → bỏ run có ts_start < since.
  3. Đọc các file stage có (01-planner.json, 02-context.json, 04-builder.json, 05-review.json). Stage thiếu → bỏ qua field tương ứng, không fail. (03-plan-review.json đã được gộp vào 02-context.json từ pipeline v4-stage.)

Nếu danh sách run sau filter = 0 → STOP No runs match filter (workspace={ws}, since={since}).

Read the full file on GitHub · 242 lines

Changes

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.

  1. yesterday First seen · 242 lines · 0 tokens per session scan A 77866ad9151b

Subscribe to this mod's changes

contextd-eval is a command published in the GitHub repository philngt/contextd (29 stars, last pushed 12d ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 2,090 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 2026-08-30.