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 skills/mplind/second-reader/second-readernpx skills add mplind/second-reader --skill second-readergit clone --depth 1 https://github.com/mplind/second-readerWhat 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 | $0.00191 | $0.06783 |
| Opus 5 | $0.00096 | $0.03392 |
| Sonnet 5 | $0.00038 | $0.01357 |
| Haiku 4.5 | $0.00019 | $0.00678 |
Grade A, and why
second-reader 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.
How it starts
The opening of the file, as written. The whole thing — 487 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Second Reader
Not a second brain. A second reader: nothing enters this vault, and nothing leaves it for the owner to act on, until an independent second reader has verified it against the sources.
What you are building
A Markdown knowledge repository, one folder that Obsidian opens as a vault, where raw source material becomes distilled, atomic, densely cross-linked, and fully sourced knowledge. The owner asks it questions and gets answers traced to their sources. It compounds: every question answered and every source read is filed back, so the vault gets richer with use.
Two properties separate this from a pile of notes, and both are non-negotiable.
Coverage. A reader of the vault should not need to have read the source. All the core, durable, useful information from a source lives in the vault, cited to its exact location. A shallow summary fails this bar.
Trust. Every claim traces to a source. Nothing is invented, nothing is over-firmed, interpretation is labelled as interpretation. The owner uses this vault to make real decisions and to speak with authority, so a fluent wrong answer is worse than an honest gap.
You hold both properties with one mechanism, described next. It is the heart of this skill.
The one rule: build, then independently validate, then loop
Never ship work from a single pass. A single pass is where defects enter, and the pass cannot see its own blind spots. Instead:
- Build the note, page, or answer in one pass (the digest).
- Validate it with a separate, independent pass that did not do the build, re-reading the source cold and auditing the build against the bar.
- Loop. If validation finds gaps or defects, run the build again with the validator's specific findings as targeted instructions, then validate again. Repeat until a validation round finds nothing.
The loop stops when a round finds nothing, not when the last batch of corrections has been made. The correction pass is itself unvalidated, and fix passes are where new defects sneak in, so a clean round after the last fix is what proves you are done.
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.
- yesterday First seen · 487 lines · 191 tokens per session scan A 74d89cee1592
second-reader is a skill published in the GitHub repository mplind/second-reader (1 stars, last pushed 4d ago), licensed MIT. It adds 191 tokens to every session and 6,783 once invoked, about $0.0010 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
design-mcp-server
Design the tool surface, resources, and service layer for a new MCP server. Use when starting a new server, planning a major feature expansion, or when the user describes a domain/API they want to expose via MCP. Produces a design doc at docs/design.md that drives implementation.
add-tool
Scaffold a new MCP tool definition. Use when the user asks to add a tool, create a new tool, or implement a new capability for the server.
api-linter
MCP definition linter rules reference. Use when bun run lint:mcp or bun run devcheck reports a lint error or warning (format-parity, schema-is-object, name-format, server-json-, etc.) and you need to understand the rule, its severity, and how to fix it. Every rule ID the linter emits has an entry in this doc.
api-context
Canonical reference for the unified Context object passed to every tool and resource handler in @cyanheads/mcp-ts-core. Covers the full interface, its RequestContext base, all sub-APIs (ctx.log, ctx.state, ctx.requestInput, ctx.inputs, ctx.enrich, ctx.content), and when to use each.
api-canvas
DataCanvas primitive reference — a Tier 3 SQL/analytical workspace for tabular MCP servers, backed by DuckDB. Use when registering tables from upstream APIs, running ad-hoc SQL across them, and exporting results. Covers the acquire → register → query → export flow, per-table TTL, the token-sharing pattern for…
api-errors
McpError constructor, JsonRpcErrorCode reference, and error handling patterns for @cyanheads/mcp-ts-core. Use when looking up error codes, understanding where errors should be thrown vs. caught, or using ErrorHandler.tryCatch in services.