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 skills add lttr/claude-marketplace --skill changeloggit clone --depth 1 https://github.com/lttr/claude-marketplaceWrote 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/lttr/claude-marketplace/changelog)<a href="https://agentmods.dev/skills/lttr/claude-marketplace/changelog"><img src="https://agentmods.dev/badge/skills/lttr/claude-marketplace/changelog/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/lttr/claude-marketplace/changelog"><img src="https://agentmods.dev/badge/skills/lttr/claude-marketplace/changelog.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 | $0.00051 | $0.01408 |
| Opus 5 | $0.00026 | $0.00704 |
| Sonnet 5 | $0.00010 | $0.00282 |
| Haiku 4.5 | $0.00005 | $0.00141 |
Grade A, and why
changelog 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 12d 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 — 99 lines — stays where its author put it; the contents beside it link to each section on GitHub.
CC Changelog
Show recent Claude Code changes, scored by relevance to the user's installed skills, commands, plugins, hooks, and usage patterns.
Workflow
- Run the data collection script:
${CLAUDE_SKILL_DIR}/scripts/cc-changelog.ts 2>/dev/null
To force a date-based lookback (ignoring lastVersion state), pass --lookback-days=N:
${CLAUDE_SKILL_DIR}/scripts/cc-changelog.ts --lookback-days=14 2>/dev/null
Use --lookback-days when the user asks for changes from a specific time period (e.g. "last 14 days", "last week" = 7 days).
- Parse the JSON output and present a concise, themed summary:
Format rules
- Hard cap: 10 bullets. The whole report is a flat, ranked list of at most 10 items — no matter how many candidates the script returns. A window with 200 entries still gets 10 bullets. Never scale the count to the input size; "there was a lot this week" is not a reason to print more. Score each candidate by combined weight (high → low):
- Personal relevance — matches user's installed skills, commands, plugins, hooks, usage patterns
- User-facing surface changes — visible message/UI text, prompts, output format, defaults the user reads or types
- Parameter / flag / config / API / schema changes — anything the user invokes or configures (CLI flags, settings keys, hook events, tool schemas, command args)
- New features / capabilities — net-new things user can do
- Behavioral changes — different output for same input, removed/changed defaults, breaking changes
- Internal / perf / refactor — deprioritize unless user-observable. A perf win that doesn't change behavior or UX ranks low. Sort by this combined score and keep only the top 10.
- Drop cosmetic items and routine bug fixes entirely (before ranking). Skip typo/wording tweaks, UI polish, and fixes for bugs the user never hit. Include a bug fix only with strong evidence it affects the user — e.g. their CLAUDE.md/settings/hooks mention a workaround for it, or it breaks a feature they demonstrably use.
- Drop what the user's settings rule out. Check
userContext.settingsfirst: a disabled feature (e.g.disableRemoteControl) means every item about it scores zero. Same for other platforms — Windows, VS Code, self-hosted runner, unused git hosts. - Merge near-duplicates into one bullet. Several fixes to the same surface (permission dialogs, MCP connections, TUI redraw) are one line, not one line each. Lead with the strongest and fold the rest in, or drop them.
- No duplicates. Each item appears exactly once. Before finishing, reread the output and remove any repeated bullet.
- Flat list, no theme headings. At 10 bullets, grouping is overhead — rank hardest-hitting first and let the list run.
- Each item = 1 short line. Rewrite verbose changelog entries into punchy summaries (5-15 words). Append version+date tag:
(2.1.47, Feb 18) - Lead with a 1-2 sentence TL;DR of the most impactful changes for this user.
- Explain WHY it's relevant by referencing what the user uses. Don't just say "matches: Bash" - say things like "you use hooks" or "affects your StatusLine setup".
- Offer the rest, don't print it. Close with one line giving the skipped count and noting you can show the second tier on request. Produce that second tier only if the user actually asks — and cap it at 15 bullets too.
What ships with it
2 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 12d ago First seen · 99 lines · 51 tokens per session scan A 5ecd42703f87
changelog is a skill published in the GitHub repository lttr/claude-marketplace (2 stars, last pushed yesterday), licensed MIT. It adds 51 tokens to every session and 1,408 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 2026-08-31.
Other skills, from other repositories
stream-chain
Stream-JSON chaining for multi-agent pipelines, data transformation, and sequential workflows.
hook-factory
Generate production-ready Claude Code hooks with interactive Q&A, automated installation, and enhanced validation. Supports 10 templates across 7 event types for comprehensive workflow automation.
html-ppt-hermes-cyber-terminal
OpenDesign + BYOK: choosing and wiring your own model, hands-on — cost, quality, and the routing decision. Built as a decision-grade AI literacy deck for engineers, IT, applied-AI teams.
r3f-best-practices
React Three Fiber (R3F) and Poimandres ecosystem best practices. Use when writing, reviewing, or optimizing R3F code. Triggers on tasks involving @react-three/fiber, @react-three/drei, zustand, @react-three/postprocessing, @react-three/rapier, or leva.
log-error-digest
Analyze log files to troubleshoot errors, identify peak error periods, and produce error clustering, frequency statistics, and time distribution reports. Supports JSON, syslog, and Nginx formats with automatic detection. Use when a user uploads a .log file and asks to analyze errors, find patterns, debug issues, or…
development
An index of programming guidance for Python, Go, Rust, TypeScript, Java, C++, and shell scripting.