Your agent records what you ship; FoxPlug drafts launch posts and a changelog. Nothing posts.
Scored across 11 tools
log_update and record_work both capture what the agent just built and are easy to confuse; the descriptions hint at a distinction (per-change update with user_facing flag vs. end-of-session milestone), but an agent could reasonably pick either. check_drift, get_plan_status, and get_brief also sit in overlapping territory, though their descriptions do a decent job of separating judgment, cheap state, and editorial output.
Names are consistently snake_case and mostly follow verb_noun (check_drift, get_brief, list_drafts, log_update, record_work, set_plan, supersede_fact). Minor deviations: the bare verb 'recall' and the interrogative phrase 'what_nobody_knows' break the pattern, but overall it reads as one coherent convention.
Eleven tools is well within the sweet spot and each maps to a distinct stage of the founder loop: set plan, log work, check drift, brief, drafts, memory, simulation, silent-shipping. Nothing feels padded or missing at the count level.
The read side is thorough (plan status, drift, brief, drafts, memory, unposted milestones), but the surface is largely read-only: there is no way to build/generate the brief, create or publish a draft, or write a new memory fact (recall is read-only and supersede_fact only retires facts). These are stated design choices, but they leave agents at dead ends for several natural workflows.
Check whether a project's recent work still matches its current plan. FoxPlug compares the plan you set (via set_plan) against the work recorded since, and returns an alignment verdict with any divergences — work that is off-plan, or plan items nothing has moved toward. Read-only judgment; it never posts anything. Free.
Inputs: project (required).
This week's brief from FoxPlug: one thing to fix, one to say and one to skip, with where each number came from. Ask it when the founder wonders what to post. Read-only. It does not rebuild the brief; the founder presses Build the brief in FoxPlug.
Inputs: project.
Get where a project stands: its current plan, how much work has been recorded since the plan was set, and the latest drift verdict (if any). Read-only and cheap — it runs no judgment, so use it to pull current state without re-running check_drift. It never posts anything. Free.
Inputs: project (required).
The latest post drafts FoxPlug wrote for the founder (channel, status, text). Read-only; nothing is posted or changed.
Inputs: limit, project.
Tell FoxPlug what you just finished building, in one short plain paragraph. Call it after a change a user would notice (user_facing true) or after plumbing nobody outside needs to hear about (user_facing false, which is kept out of drafts and counts toward "not worth a post"). It never posts anything. Free. Do NOT include secrets, keys, or tokens.
Inputs: summary (required), details, project, user_facing.
What FoxPlug remembers about this maker, with an epistemic status on every fact: settled (more than one observation, none contradicting), thin (seen once), contested (the evidence disagrees with itself), superseded (replaced by something later — never propose it again). Read-only. Free.
Inputs: project.
Record a build-work summary into FoxPlug as a milestone (source: Claude Code). Call this at the end of a session. It captures what you shipped so FoxPlug can score it and surface the shippable moments in your Radar. It never posts anything. Free — capture is not metered. Do NOT include secrets, keys, or tokens in the summary.
Inputs: project (required), summary (required), decisions, files_touched.
Set the current plan, intent, or priorities for a project in FoxPlug (source: Claude Code). Call this when you set or change what a project is working toward, so FoxPlug knows the goal your later work is measured against. It replaces the project's previous current plan. Free — it never posts anything. Do NOT include secrets, keys, or tokens in the plan.
Inputs: plan (required), project (required), horizon.
Before you merge or ship, ask your buyers. Give the change as the buyer would see it (pricing text, a feature description, launch copy). FoxPlug builds 3 to 5 buyer personas ONLY from evidence the founder already gave it (their site, the feedback requests they connected, what they shipped), each citing its evidence, and returns how each would react: would buy, hesitate or leave, the objection, the exact sentence that confused them and what would change their mind, plus the top objections and a tighter rewrite. It is a SIMULATION, not real customers, and says so. The first simulation per project is free; more need a paid plan. Nothing is posted or changed.
Inputs: change (required), project.
Mark a remembered fact as finished, so it is never proposed again. Takes the fact id from recall and a REASON — a fact struck out without a reason cannot be restored. This is the only one of these tools that changes anything, and it changes memory, never anything public. Free.
Inputs: reason (required), fact_id (required), replacement_fact_id.
What this project shipped that nobody outside it knows about. Counts the milestones worth talking about in a window and how many of them were actually posted, and names the loudest silent ones. Read-only — it generates nothing and posts nothing. Free.
Inputs: project (required), days.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Release notes and changelogs from across the web, indexed for agents. Also via CLI and REST API.
Publish and update your startup on BetaFinds from an AI agent — listings, changelog, releases.
Ask what shipped, and draft, edit and approve launch posts, changelogs and docs updates.
Generate changelogs, release emails, help-center articles, banners, and social posts from commits.
Add one secure layer between your agents and this server.
