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 gtapps/claude-code-hermit/plugin install feed-hermitWrote 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/gtapps/claude-code-hermit/feed-brief)<a href="https://agentmods.dev/skills/gtapps/claude-code-hermit/feed-brief"><img src="https://agentmods.dev/badge/skills/gtapps/claude-code-hermit/feed-brief/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/skills/gtapps/claude-code-hermit/feed-brief"><img src="https://agentmods.dev/badge/skills/gtapps/claude-code-hermit/feed-brief.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 | $0.00104 | $0.03438 |
| Opus 5.5 | $0.00042 | $0.01375 |
| Sonnet 5.5 | $0.00021 | $0.00688 |
| Haiku 4.5 | $0.00010 | $0.00344 |
Grade A, and why
feed-brief 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.
Copies of this mod
1 near-identical copy found in the catalogue:
- feed-brief — 89% identical, 49 lines differ
How it starts
The opening of the file, as written. The whole thing — 229 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Feed Brief
Run the full 7-phase brief pipeline for this project. Config lives at the project root
(feed-sources.md, feed-categories.md, FEEDS.md); state and archive live under .claude-code-hermit/.
Flags
--morning: slotmorning. Forward-looking. Emphasize what's new since yesterday evening, what's worth opening today, emerging patterns. Tone: calm, sharp, selective.--evening: slotevening. Backward-looking. Emphasize what changed during the day, what escalated or faded, what to carry into tomorrow. Tone: slightly more reflective.--slot <name>: custom slot label (e.g.midday). Use<name>verbatim wherever<slot>appears below. Orientation defaults to the morning (forward-looking) framing unless the operator'sFEEDS.mdsays otherwise.
Resolve <slot> from the flag before starting. All filenames below use that resolved value.
Steps
- Read
FEEDS.md— brief philosophy, scoring rules, output format, and section guidance. - Read
feed-sources.md— full source list with types and URLs. - Read
feed-categories.md— operator's current focus areas and priority tiers (P1/P2). - Read
.claude-code-hermit/config.json— thefeed.enrichments.*andfeed.reaction_feedbackgates consulted in Phases 3 and 5b. A gate runs only when its value istrue; treat an absent key as disabled.
Phase 1 — Web/RSS sources
Before dispatch, run bun ${CLAUDE_PLUGIN_ROOT}/scripts/source-fetch-result.ts new-run.
Keep its returned UUID as the expected run_id for this invocation; never infer it from an
existing file or the agent's reply. If generation fails, follow the wholesale fetch-failure
path below without dispatching.
Dispatch the @feed-hermit:source-fetcher subagent (model: haiku) to fetch all web and rss
sources from feed-sources.md. Pass the full source list (URLs + names), the resolved <slot>, the absolute project root and output path, and the generated run_id to copy exactly into the output JSON.
Output-path contract: the agent writes its extracted items to tmp/feed-source-items-<slot>.json
in the project root (never /tmp/). Pass <absolute-project-root>/tmp/feed-source-items-<slot>.json to the agent so its inherited working directory cannot redirect the write. Items are nested under sources[], each item:
{title, summary, url, published_at, source, section, author}. No scoring —
extraction only. After the agent returns, run:
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 e20d9b1786c6
- 1mo ago Changed 6062bb1f1dd3
- 1mo ago Changed · +13 lines 22e9383ecbc1
- 1mo ago Changed 3e29f45d2b45
- 2mo ago First seen · 216 lines · 104 tokens per session scan A 9a8be9c377e2
feed-brief is a skill published in the GitHub repository gtapps/claude-code-hermit (83 stars, last pushed 3d ago), licensed MIT. It adds 104 tokens to every session and 3,438 once invoked, about $0.0004 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 2026-09-03.
Other skills, from other repositories
monday-planning
Suggest a weekly training structure from recent load and deliver the plan.
fresh-start
Continue in a new chat without losing anything — writes a compact handoff to paste into a fresh conversation. Use whenever the user wants to move to a new chat or conversation, hits usage limits, says replies got slow or Claude forgets earlier details, or the conversation has grown long.
token-check
Estimate how many tokens a pasted text, an attached file or the conversation so far takes, what it means for the context window and usage limits, and how to cut it. Use when the user asks how big something is in tokens, why they keep hitting limits, or whether a file is too large to send.
ccc-adhd
Answer-first output: the fix, then command, then file:line, context last. Toggle on, off or status; stacks with caveman. Based on ayghri/i-have-adhd (MIT). Use when the user wants the answer first.
ccc-resume-session
Resume a saved session file: read it fully, then brief state, blockers and next step before any work. Use when continuing earlier work.
ccc-spawn
Spin self-contained work off on demand: a background subagent, a separate session (Cowork spawntask chip or CLI /spawn peer), or an independent expert pass; routes fan-outs to $ccc-fleet and chains to $ccc-relay. Use when the user types $ccc-spawn or wants a task done outside this thread.