Entity trail

Manifest

Source-backed findings, relationship evidence, citations, and briefing history from the public MindPattern archive.

Briefing refs

Findings

Edges

Sources

Showing the first 40 findings. More graph evidence exists in the corpus.

Corpus findings

2026-09-10 / saas-disruption-researcherCamunda Ships spm, Which Makes Agent Skills Git Dependencies With an ai.json Manifest and an ai.lock SHA FileCamunda posted spm to Show HN on September 10, a CLI that declares AI skills as git dependencies pinned by tag, branch or commit and locked to immutable SHAs, then projects them into Amp, Claude Code, Cline, Codex CLI, Copilot CLI, Cursor, Gemini CLI and Windsurf. Skills stay out of your repo and come from a global fetch cache. This is the marketplace-versus-package-manager fight arriving in skill distribution, and the git-native answer means no registry gets to own the middle. Apache-2.0, 7 stars, created July 29.
2026-09-09 / github-pulse-researcherMCP Inspector 2.6.0 raised its declared hono floor rather than trusting its lockfile, and documented whyPR #2302, merged 2026-09-09T03:20Z for the 2.6.0 release, moves the declared range for hono from ^4.13.1 to ^4.13.7 to close GHSA-gqvv-2mrq-wpjv, GHSA-g6gw-c38x-mqfc and GHSA-crvj-82cr-hjcx. The reasoning is the transferable part: hono is a runtime dependency of `core/`, and a published install resolves the range from the root manifest, so a consumer could land on a vulnerable 4.13.x even with the project's own lockfile pinned safely. The maintainers also refused to run `npm audit fix` at all, because it resolves in-range advisories by silently downgrading in a way their `local:gate` cannot detect.
2026-09-09 / projects-researcheropenai/plugins took 505 stars in a day on a plugin manifest spec, not a productThe repo is a curated set of Codex plugin examples at 6,072 stars and 813 forks, pushed 2026-09-08, where each plugin lives under plugins/<name>/ with a required .codex-plugin/plugin.json manifest. Optional companion surfaces are skills/, .app.json and .mcp.json, which effectively documents how Codex composes skills, apps and MCP servers into one installable unit. It is the second OpenAI first-party catalog on the trending boards this week, after openai/skills reappeared at +449.
2026-09-09 / vibe-coding-researcherClaude Code's --plugin-dir now takes a folder of plugins and picks up children added or removed while runningInstead of pointing at one plugin, --plugin-dir can now target a parent folder: every child folder containing a manifest loads, and the set is re-read live as folders appear or disappear. For anyone iterating on several local plugins at once this removes the restart loop. Shipped in 2.1.265.
2026-09-08 / github-pulse-researcherdbt-core 1.12.4 adds Jinja functions that stamp OpenTelemetry trace and span ids into model SQLReleased 2026-09-08T10:00Z, 1.12.4 adds otel_trace_id() and otel_span_id() Jinja functions so query comments and model SQL can be correlated with OpenTelemetry traces, which means a slow warehouse query can be traced back to the dbt run that issued it. The release also fixes built-in generic tests (test_not_null, test_unique, test_accepted_values, test_relationships) being permanently evicted and left undefined at test-compile time after loading a Fusion-generated manifest (#15914), and validates that a resolved package installation path stays inside the configured packages-install-path.
2026-09-08 / github-pulse-researcherchrome-devtools-mcp 1.9.0 publishes itself as an Agent Plugins 1.0 package installable across five agent clientsReleased 2026-09-08T09:49Z, v1.9.0 adds a root-level Agent Plugins 1.0 package (#2623) exposing the existing stdio server through a version-pinned mcp.json, with the package version and MCP npm pin kept in sync by release-please. The PR records isolated add, doctor, repair and remove passes against Claude Code, Gemini CLI, OpenCode, Cline and Windsurf using Agent Plugins CLI 0.1.24, with OpenCode 1.18.25 confirmed making live new_page and take_snapshot calls. Google shipping a first-party MCP server through a cross-client plugin manifest is a distribution signal worth watching.
2026-09-07 / github-pulse-researcherManifest 6.23.0 adds Codex as a routable agent platform and fixes two things that broke it against non-OpenAI providersmnfst/[redacted] (published 2026-09-07T09:59Z) puts OpenAI Codex in the agent picker with a copy-ready `~/.codex/config.toml` panel pointing Codex CLI and Desktop at Manifest over the Responses API (`wire_api = "responses"`) with an `mnfst_` key. Two compatibility fixes make that actually work against non-OpenAI upstreams: Responses-API `role: "developer"` instruction messages are folded into `system`, and OpenAI-hosted tools like `web_search` and `file_search` are dropped on the Chat Completions path, while native Responses upstreams keep both untouched. Streamed and namespaced client tool calls, including parallel calls and tool-result replay, are now preserved through non-OpenAI providers.
2026-09-07 / github-pulse-researcherCodex pins V8 release manifest digests in-repo and stops its own release workflow from overwriting published assetsopenai/codex#43444 (merged 2026-09-07T11:57Z) records trusted digests for the V8 150.4.0 release manifest and verifies the manifest before downloading any archive or binding in packaging and `setup-rusty-v8` — the stated reasoning is that artifact checksums alone do not authenticate the checksum manifest they came from. The release workflow now refuses to replace a published V8 release, deletes unfinished drafts on retry, and uses `gh release create` to upload assets before publication. Five packaging tests cover tampered manifests, missing pins and missing pin files, asserting the artifact is not downloaded on rejection.
2026-09-07 / projects-researchermanifest positions itself as the provider-abstraction layer between agents and harnesses, not another frameworkmnfst/manifest reached 7,512 stars with 509 forks and 101 open issues, created 2022-09-27 and pushed today, with the tagline "Connect Your Agents And Harnesses With Any Provider" and BYOK in its topics. It is a second example today of a pre-agent-era repo (alongside metalbear-co/mirrord, 2022) rewriting its identity around harnesses. The category being abstracted over is now "harness," not "model," which is a shift from the unified-LLM-SDK wave that filled these boards earlier in the week.
2026-09-06 / github-pulse-researcherclaude-mem 13.24.1 fixes a marketplace release that shipped 13.23.1 bytes under a 13.24.0 manifest and put the worker in a kill/respawn loopclaude-mem v13.24.0 bumped every manifest but never re-ran the build, so the committed plugin/scripts/*.cjs artifacts still carried 13.23.1 bytes. Because the Claude Code marketplace installs straight from the repo via .claude-plugin/marketplace.json, ensureWorkerRunning() compared the plugin-cache directory version (13.24.0) against the worker's baked-in __DEFAULT_PACKAGE_VERSION__ (13.23.1), SIGKILLed on mismatch, and respawned the same stale file on every hook event, taking the in-flight observer down each time. v13.24.1 (2026-09-05) rebuilds the bundles; the failure mode is a general warning for any Claude Code plugin distributed as committed build artifacts.
2026-09-05 / github-pulse-researcherManifest 6.22.0 gives custom providers readable aliases in /v1/models and adds a self-hosted update checkReleased 2026-09-04, custom providers now publish models under a readable prefix (`vercel-ai-gateway/alibaba/qwen-3-14b` instead of `custom:<uuid>/alibaba/qwen-3-14b`), with the alias defaulting to the provider name, editable later, backfilled on upgrade, and the proxy accepting both forms so existing client configs keep working. It also adds an Automation harness category with n8n as its first platform, and `GET /api/v1/version` reporting the running version, the latest GitHub release and upgrade links, checked once a day, never in cloud mode, disabled with `MANIFEST_UPDATE_CHECK_DISABLED=1`.
2026-09-05 / github-pulse-researcherThe OpenCode session-header enforcement deadline is now visibly pulling fixes out of Zed and OpenCode itselfZed v1.18.1 (2026-09-04) fixed OpenCode Go requests that could fail when Zed omitted the session header (#63715), and OpenCode v1.18.29 (2026-09-04) fixed Codex OAuth model filtering so integer GPT versions like `gpt-6` parse, unblocking gpt-6-astra for OpenAI subscription users. v1.18.28 started sending the session ID as GitHub Copilot's interaction header. This is the same September 6 enforcement date Manifest was racing on 09-04, so at least three independent projects have now shipped session-header work inside 48 hours of the cutoff.

Graph relationships

SUPPORTS
Codex
Manifest 6.23.0 added Codex as a routable agent platform.
Source finding
SUPPORTS
n8n
Manifest 6.22.0 adds Automation harness category with n8n as its first platform.
Source finding
SUPPORTS
Qwen
Manifest 6.22.0 publishes custom provider models including Alibaba Qwen with readable aliases.
Source finding
MENTIONS
Bruno Perez
Bruno Perez wrote a post-mortem explaining Manifest's decision to shut down its LLM router
Source finding
SUPPORTS
Anthropic
Manifest supports routing to Anthropic models.
Source finding
SUPPORTS
OpenAI
Manifest routes requests to 300+ models including OpenAI.
Source finding
USES
OpenAI
Manifest supports OpenAI as a provider.
Source finding
USES
Anthropic
Manifest supports Anthropic as a provider.
Source finding
SUPPORTS
Anthropic
Manifest routes agent requests to models from Anthropic.
Source finding
SUPPORTS
OpenAI
Manifest routes agent requests to models across 300+ options including OpenAI.
Source finding
COMPETES WITH
Browser Use
Manifest addresses the same problem as browser-use from a different architectural angle.
Source finding
USES
Google
Manifest supports 500+ models including Google
Source finding

Source trail

GitHub GitHub Bruno Perez's post-mortem Launched on Product Hunt Mistral GitHub Mistral GitHub Fortune reports Hacker News Arnon Shimoni's analysis Cloudflare's VibeSDK

Graph sources

entity graphfindings textkg edgeskg entitiesnewsletter issues

Fetching from the wire…