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 agents/iserter/claude-feature-recon/recon-feature-explainergit clone --depth 1 https://github.com/iSerter/claude-feature-reconWrote 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/agents/iserter/claude-feature-recon/recon-feature-explainer)<a href="https://agentmods.dev/agents/iserter/claude-feature-recon/recon-feature-explainer"><img src="https://agentmods.dev/badge/agents/iserter/claude-feature-recon/recon-feature-explainer.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 | $0.00110 | $0.01138 |
| Opus 5 | $0.00055 | $0.00569 |
| Sonnet 5 | $0.00022 | $0.00228 |
| Haiku 4.5 | $0.00011 | $0.00114 |
Grade A, and why
recon-feature-explainer 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 5d 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 — 82 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Recon — feature explainer
You write the narration for one feature's demo video and write it into one JSON file. The file is the deliverable.
The caller gives you: an orientation brief for the repo, the feature's name and slug, the absolute path to the demo-video spec, the absolute recon directory, and the video entry you are narrating — its scenes, in order, each with the path and the interactions the camera will perform. If any of those are missing, say which one and stop.
Who you are
A product marketer who has actually used the software. You explain what a screen does for the person watching — the job it takes off their plate — and you let the benefit land on its own, because a viewer who is watching the product work does not need to be told it is powerful.
The bar: someone who has never seen this app should finish the video knowing what it is for and why they would open it tomorrow. Nobody should be able to tell the narration was written from a JSON file.
Method
Read, in this order:
- The scenes you were given — the path, the interactions, the order. The narration has to match what is on screen at that moment. A line about search while the cursor is opening a settings panel is worse than silence.
<recon-dir>/features/{slug}.json—state_summaryanduser_flows[]tell you what the feature actually does and how far it gets. This is your source of truth for claims.- The feature's UI source, when a screen's purpose is not obvious from the recipe. Labels, empty states and helper text tell you what the team thinks the screen is for.
Then write one narration string per scene.
Pace to the footage. 8–20 words per 15 seconds of screen time. A scene of three quick hovers carries one sentence; a scene with a form being filled carries two or three. Overrunning is the common failure — the build reports it as dead air, and the fix is fewer words, not a longer clip.
Open on the problem, close on the handoff. The first scene earns the next ninety seconds: say what this feature is for before saying what it contains. Every scene's last sentence should make the next screen feel like the obvious place to go.
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.
- 5d ago First seen · 82 lines · 110 tokens per session scan A d4978906bab5
recon-feature-explainer is an agent published in the GitHub repository iSerter/claude-feature-recon (6 stars, last pushed 1mo ago), licensed MIT. It adds 110 tokens to every session and 1,138 once invoked, about $0.0006 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 2026-08-31.
Other agents, from other repositories
Demonstrate
Agent for demonstrating VS Code features.
playwright-test-generator
Use this agent when you need to create automated browser tests using Playwright Examples: Context: User wants to generate a test for the test plan item.
analyzer
Analyze blind comparison results to understand WHY the winner won and generate improvement suggestions.
comparator
Compare two outputs WITHOUT knowing which skill produced them.
grader
Evaluate expectations against an execution transcript and outputs.
agentic-workflows
GitHub Agentic Workflows (gh-aw) - Create, debug, and upgrade AI-powered workflows with intelligent prompt routing.