{"id":19887,"plugin_id":"plugins_6aa164d9f84881919eb4d5524bf952ac","kind":"skill","collection_source":null,"comparison_source":null,"observed_at":"2026-09-30T23:15:55.482Z","digest":"445cafb2db2dc226d1712604b4c1d8efd4702ae83782fac3aebddfbda17c5ea0","against":null,"payload":{"name":"survey","description":"Investigate a topic across papers, articles, social-media posts, and industry signals, then deliver an indexed Markdown report. Use when the user asks to investigate, survey, gather sources, build an index, collect evidence, round up references, or otherwise wants a thorough cross-source roundup on a specific topic. Depends on documenting-with-sources and writing-quotation.","included_files":[],"skill_md_contents":"---\nname: survey\ndescription: Investigate a topic across papers, articles, social-media posts, and industry signals, then deliver an indexed Markdown report. Use when the user asks to investigate, survey, gather sources, build an index, collect evidence, round up references, or otherwise wants a thorough cross-source roundup on a specific topic. Depends on documenting-with-sources and writing-quotation.\n---\n\n# Survey\n\nGather sources from the web, papers, social media, and industry on a specific topic, and produce an indexed Markdown report. First read `../../references/runtime-capabilities.md`.\n\nThis skill follows the shared sourced-writing conventions defined in `documenting-with-sources`. Read `documenting-with-sources` before drafting.\n\n## Quality criteria\n\nApply to every source.\n\n- Prefer trustworthy sources (peer-reviewed papers > official blogs > major industry media > personal blogs).\n- If the author is an individual, list their affiliation and role. If unknown, look it up.\n- For papers, in addition to bibliographic info (authors, affiliations, venue, year), include citation count.\n- Always attach a URL.\n- Always attach a date.\n- Treat official documentation and third-party articles as different reliability tiers. Do not mix them; the reader must be able to tell which is which.\n\n## Prose structure: Assertion-Evidence form\n\nWrite the body in Assertion-Evidence form — claim first, then evidence.\n\n1. Prose: the writer states the claim or summary in their own words first.\n2. Immediately after, a code-block quotation from the source backs up the claim, formatted per `writing-quotation`.\n3. On the line after the closing fence, place the source reference `[source-name (YYYY/MM)]`.\n\nThe reader grasps \"what is being said\" first, then checks \"what is the basis\". The reverse order (quotation first, claim later) is forbidden — the reader cannot tell what the quotation is for until they have read past it.\n\nBad example (quotation first):\n\n```\nThe standard recipe of pretraining on huge corpora and then running classical preference-label RLHF is now widely treated as obsolete.\n```\n\n[industry-tracker (2026/03)]\n\nThe mainstream has shifted to a modular post-training stack.\n\nGood example (claim first):\n\nThe classical RLHF pipeline (human preference labels → reward model → PPO) is no longer used in leading models; it has been replaced by a modular stack that separates concerns.\n\n```\nThe standard recipe of pretraining on huge corpora and then running classical preference-label RLHF is now widely treated as obsolete. Every leading model released in the past year uses a different post-training stack.\n```\n\n[industry-tracker (2026/03)]\n\n## Source-reference label\n\nWithin the `[label (YYYY/MM), location]` structure defined in `documenting-with-sources`, the survey skill fills the label slot with the publication or source name (media name, site name, etc.). The location is omitted when it cannot be pinned down.\n\n## Heading-content alignment\n\nSection headings and the items placed under them must match exactly.\n\n- If a heading is \"human-side guardrails\", only human conduct and discipline goes underneath. Tooling and CI/CD belong under a separate heading.\n- If a heading is \"failure cases\", do not mix in success stories or recommendations.\n- If a source spans multiple angles, either split it across the relevant sections, or place it under the most appropriate one and add an explicit note about the other angles.\n- Before finalising, walk every heading and check that everything underneath it actually belongs there.\n\n## Output destination\n\nWrite the deliverable to `{CWD}/reports/` as a `.md` file. Create the directory if it does not exist. Sub-agents that emit intermediate artefacts use the same directory.\n\n## Workflow\n\n1. From the user's request, identify the claim or hypothesis and the collection scope.\n2. Design search queries for the scope (in multiple languages where appropriate).\n3. Use parallel subagents by angle when available; otherwise investigate the identical angles sequentially with the same checklist.\n4. Consolidate the results; the main agent assembles the final version.\n5. Write out the deliverable to `{CWD}/reports/` as a `.md` file.\n6. Record `audit_mode: parallel-subagents` or `audit_mode: sequential-single-agent`, completed checks, and limitations.\n\n## Sub-agent delegation rules\n\nWhen delegating to sub-agents, follow these. If unavailable, apply every item sequentially; do not skip checks.\n\n- Split by angle and dispatch in parallel (e.g. \"papers & academia\", \"media & blogs\", \"social media\", \"industry signals\").\n- Give each sub-agent the quality criteria and the conventions from `documenting-with-sources`. Re-emphasise \"no fabricated associations or interpretations\" specifically — sub-agents are particularly prone to drifting toward the calling-conversation context and inventing connections.\n- Sub-agent output is not the final deliverable. The main agent performs:\n  - Deduplication.\n  - Information completion (filling in missing affiliations, citation counts, etc. via additional lookups).\n  - Structural unification (tables, consistent section structure).\n  - Separation of criticism from supporting evidence.\n  - Explicit listing of investigation limits (information that could not be retrieved, unverified URLs, etc.).\n  - Removal of fabricated associations (see the corresponding section in `documenting-with-sources`).\n  - Heading-content alignment check (see above).\n\n## Deliverable structure\n\n```\n# Survey of {topic}\n\nDate: YYYY-MM-DD\nScope: {scope description}\n\n## Table of Contents\n\n1. [{angle 1}](#1-angle-1-slug)\n2. [{angle 2}](#2-angle-2-slug)\n...\nN. [Criticism & concerns](#n-criticism-concerns)\nN+1. [Overall assessment](#n1-overall-assessment)\nN+2. [Investigation limits](#n2-investigation-limits)\n\n## 1. {angle 1} (e.g. academic papers)\n## 2. {angle 2} (e.g. media coverage)\n## 3. {angle 3} (e.g. social-media reactions)\n## 4. {angle 4} (e.g. industry signals)\n## N. Criticism & concerns\n## N+1. Overall assessment\n## N+2. Investigation limits\n```\n\nAdjust the section layout for the topic.\n\n## Table-of-contents requirements\n\nAlways place a table of contents at the top of the report, directly after the metadata block and before the body. Writing the report without a ToC is forbidden.\n\n- If there are ten or more sections, or fifteen or more individual items (papers, articles, etc.), use a two-level ToC. Level 1 is the section name; level 2 is the section's main items (paper titles, article headlines, ...).\n- For short reports (five or fewer sections, few items per section) a single-level ToC is fine.\n- Provide anchor links. Use the renderer's slug rules (lowercased, spaces → hyphens, special characters dropped) for the link targets.\n- Even when the slug rule is uncertain, write the link rather than dropping it; let the renderer slugify.\n- ToC entries and section headings must match word for word. No abbreviation or paraphrase.\n"},"changes":[],"summary":"First saved snapshot. No earlier version is available for comparison.","summary_kind":"deterministic","summary_metadata":{}}