@companion-ai/feynman 0.5.63 → 0.5.64

package/RELEASES.md

@@ -6,6 +6,10 @@ GitHub release notes are generated from the matching `## vX.Y.Z` section in this
 
 ## Unreleased
 
+## v0.5.64 - 2026-11-05
+
+- **pi-subagents 0.85.0 and pi-web-access 0.47.0.** Workflow scripts that launch parallel researchers now run in a locked-down process, a research prompt with long runs of separators or `git` options no longer stalls a workflow, and resuming a background run no longer trusts files that run could rewrite (runs started before this update must be relaunched). Fetching a GitHub repository now returns READMEs up to 100K characters instead of 8K, and on Windows cloned repositories go to your temp folder instead of `C:\tmp`. `feynman search set` now also accepts the new `sofya` and `outcrawl` search providers, which are used only when you pick them. `/lit` now launches parallel researchers with the same workflow form as `/deepresearch`, fixing a first-call `parallel is not defined` failure.
+
 ## v0.5.63 - 2026-11-04
 
 - **pi-btw 0.8.1.** While a `/btw` side question is being answered, messages you submit in the main editor are no longer held until it finishes; the answer streams in the overlay. Side threads also no longer repeat the working directory section of the system prompt.

package/dist/pi/web-access.js

@@ -9,7 +9,7 @@ export const PI_WEB_SEARCH_PROVIDERS = [
     "tavily", "you", "firecrawl", "jina", "searxng", "degoog", "duckduckgo", "perplexity", "gemini", "kimi", "exa",
     "serpdive", "kagi", "ollama", "anysearch", "xai", "mistral", "brightdata", "serpbase", "serpapi", "serper", "serply",
     "valyu", "bocha", "xcrawl", "baizhi", "zai", "keenable", "ceramic", "deepseek", "linkup", "ninerouter",
-    "cloudflare", "searchapi", "tabstack",
+    "cloudflare", "searchapi", "outcrawl", "sofya", "tabstack",
 ];
 // The web-search.json field holding a provider's key (searxng and ninerouter take their base URL);
 // undefined for providers with nothing to set.

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@companion-ai/feynman",
-  "version": "0.5.63",
+  "version": "0.5.64",
   "description": "The open source AI research agent.",
   "license": "MIT",
   "type": "module",
@@ -85,8 +85,8 @@
     "fast-xml-parser": "5.11.2",
     "pi-btw": "0.8.1",
     "pi-docparser": "4.0.0",
-    "pi-subagents": "0.84.0",
-    "pi-web-access": "0.46.0",
+    "pi-subagents": "0.85.0",
+    "pi-web-access": "0.47.0",
     "posthog-node": "^5.53.0",
     "undici": "^8.12.3"
   },

package/prompts/lit.md

@@ -12,7 +12,15 @@ Derive a short slug from the topic (lowercase, hyphens, no filler words, ≤5 wo
 
 1. **Plan** — Outline the scope: key questions, source types to search (papers, web, repos), time period, expected sections, and a small task ledger plus verification log. When the input appears to name a lab, PI, author, institution lab page, or author profile, run the review as a publication-corpus review: find the lab/author identity first, collect the reachable publication list, then map the research trajectory across that corpus. Write the plan to `outputs/.plans/<slug>.md`. Briefly summarize the plan to the user and continue immediately in the same turn with the next tool call; never end a turn with a message that only announces what you will do next. Do not ask for confirmation or wait for a proceed response unless the user explicitly requested plan review.
    - When updating the plan ledger later, keep edits small and valid. If an `edit` tool call fails with a JSON parse error or the replacement would require embedding a large markdown block, rewrite the full corrected plan file with the file-writing tool instead, then continue to final artifact/provenance verification.
-2. **Gather** — Use the `researcher` subagent when the sweep is wide enough to benefit from delegated paper triage before synthesis. For narrow topics, search directly. Researcher outputs go to `outputs/.drafts/<slug>-research-*.md`; pass that path as each researcher's `output`. For publication-corpus reviews, the lead agent owns identity resolution and writes `notes/<slug>-publications.md` with reachable titles, years, venues, URLs/DOIs, and gaps before delegating trajectory synthesis. Prefer lab publication pages, author profiles, arXiv/OpenReview/Semantic Scholar pages, and paper search results that expose stable source URLs. Do not silently skip assigned questions; mark them `done`, `blocked`, or `superseded`.
+2. **Gather** — Use the `researcher` subagent when the sweep is wide enough to benefit from delegated paper triage before synthesis. For narrow topics, search directly. Researcher outputs go to `outputs/.drafts/<slug>-research-*.md`. For publication-corpus reviews, the lead agent owns identity resolution and writes `notes/<slug>-publications.md` with reachable titles, years, venues, URLs/DOIs, and gaps before delegating trajectory synthesis. Prefer lab publication pages, author profiles, arXiv/OpenReview/Semantic Scholar pages, and paper search results that expose stable source URLs. Do not silently skip assigned questions; mark them `done`, `blocked`, or `superseded`.
+   - For parallel researchers, write a workflow script that returns `await runs.all([...])` to `outputs/.plans/<slug>-workflow.js` with the write tool. In a later tool call, run `subagent` with `{ "workflow": "./outputs/.plans/<slug>-workflow.js", "async": true, "globalConcurrencyLimit": 4 }`. Each item has a unique `key`, an `agent`, and a short `task` naming its output path; items take no `output` field. Only `runs.all` is available; there is no `parallel()`.
+
+     ```js
+     return await runs.all([
+       { key: "foundations", agent: "researcher", task: "Read outputs/.plans/<slug>.md and write outputs/.drafts/<slug>-research-foundations.md." },
+       { key: "recent", agent: "researcher", task: "Read outputs/.plans/<slug>.md and write outputs/.drafts/<slug>-research-recent.md." },
+     ]);
+     ```
 3. **Synthesize** — Separate consensus, disagreements, and open questions. For publication-corpus reviews, also identify 3-5 research trajectories and the 3-5 papers that most changed the corpus direction; rank them by contrastive originality, methodology strength, and relationship to prior art rather than by author prestige alone. When useful, propose concrete next experiments or follow-up reading. Use Mermaid diagrams for taxonomies, method pipelines, or lab trajectory maps when the structure is source-supported and changes the reader's research decision. Keep the output to research evidence, source coverage, and next research decisions; do not create non-research operational artifacts from a literature review run.
 4. **Cite** — Spawn the `verifier` agent to add inline citations and verify every source URL in the draft.
 5. **Verify** — Spawn the `reviewer` agent to check the cited draft for unsupported claims, logical gaps, zombie sections, and single-source critical findings. Fix FATAL issues before delivering. Note MAJOR issues in Open Questions. If FATAL issues were found, run one more verification pass after the fixes.