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 iliaal/codesage/plugin install codesage-toolsWrote 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/iliaal/codesage/codesage-report)<a href="https://agentmods.dev/commands/iliaal/codesage/codesage-report"><img src="https://agentmods.dev/badge/commands/iliaal/codesage/codesage-report/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/commands/iliaal/codesage/codesage-report"><img src="https://agentmods.dev/badge/commands/iliaal/codesage/codesage-report.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 | $0.00018 | $0.00684 |
| Opus 5.5 | $0.00007 | $0.00274 |
| Sonnet 5.5 | $0.00004 | $0.00137 |
| Haiku 4.5 | $0.00002 | $0.00068 |
Grade A, and why
codesage-report scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 1mo ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 32 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Render CodeSage findings
Run the packaged formatter with the user's arguments as distinct shell arguments:
"${CLAUDE_PLUGIN_ROOT}/bin/codesage-review-state" report "$PROJECT"
Require an absolute project path. Pass optional --status, --severity, --category, --feature, and --output / -o directly. Do not generate, summarize, or rewrite the Markdown with a model. Return the helper's stdout verbatim; if it fails, report the error and do not fabricate a report.
Filters intersect. Defaults are statuses open,wont-fix, all three severities, and all four categories. Unknown filter values fail. Without --output, Markdown goes to stdout. With an output path, the helper atomically writes UTF-8 Markdown and prints the destination, selected finding count, and affected feature count. The destination's parent must already exist; symlink components and nonregular destinations are refused.
Persisted state and coverage
Read title, kind, entry_path, and feature_files from findings documents. For legacy metadata gaps, the formatter makes one complete CLI inventory lookup (features-list --json --limit 0) and can join records by feature_id. Missing IDs or a failed lookup display (metadata unavailable); never guess an entry from a finding's location. For repeatable archival rendering, pass --features <inventory.json> from the inventory helper instead of a live lookup. An explicitly supplied malformed or truncated inventory fails the report.
The helper checks .codesage path components and every findings JSON leaf, rejects symlinks and nonregular files, and fails on malformed state. It uses the latest persisted reviewed_at for the state timestamp, never the wall clock. Identical findings, metadata inventory, and source contents with identical arguments produce byte-identical output. Source contents matter because live acknowledgement diagnostics check whether cited files changed; these diagnostics can change even when findings JSON does not.
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.
- 1mo ago Changed · -104 lines · -6 tokens per session 4f9df50fde18
- 1mo ago First seen · 136 lines · 24 tokens per session scan A 2acc437bf85b
codesage-report is a command published in the GitHub repository iliaal/codesage (21 stars, last pushed yesterday), licensed MIT. It adds 18 tokens to every session and 684 once invoked, about $0.0001 per session on Opus 5.5. 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
call-graph
Trace callers, callees, or paths using the call graph.
peek
Quickly find likely code locations without returning full code.
search
Search codebase by meaning using semantic search.
visualize
Generate temporal HTML call graph visualization.
index
Index the codebase for semantic search.
reindex
Fully rebuild the codebase index from scratch.