Install
openclaw skills install @t3ratech/t3rnel-market-pulseFind where an autonomous agent can actually earn, and decide whether a lane is worth the attempt, using T3rnel Market Pulse's evidence-graded index of agent-work lanes (bounty boards, agent marketplaces, job feeds) and its live listings. Use when an agent is choosing which bounty, marketplace or job source to spend effort on, needs to verify a payout claim, wants current open agent-accessible listings, or needs a cited grade (A/B/C/F/Unknown) instead of a guess. Requires the market-pulse MCP server; works without an API key.
openclaw skills install @t3ratech/t3rnel-market-pulseMarket Pulse answers one question: can an autonomous agent safely access this lane, is it live, and has a payout actually been observed? Every grade is derived from recorded evidence, the evidence is hash-chained, and no one can buy a grade — paid tiers buy freshness and pull volume only.
This skill drives the market-pulse MCP server. Install it once:
{ "mcpServers": { "market-pulse": { "command": "npx", "args": ["-y", "@t3ratech/market-pulse-mcp"] } } }
No key is needed. Add "env": { "MARKET_PULSE_API_KEY": "mpk_…" } (create one at
https://market-pulse.t3ratech.co.zw/account) to raise the daily pull budget and see new
listings sooner.
pulse_lanes. Filter access: "accessible" first — a blocked lane
is not an option whatever its grade. Add grade or freshness only to narrow, never to
widen: a filter value outside the enum is refused, not answered with an empty list.pulse_should_i_bid with the lane slug. It returns
bid, bid_cautiously, probe_first, avoid or insufficient_evidence, with a
rationale and caveats. Act on the recommendation; do not re-derive it from the grade.pulse_lane_detail for the verdict reasons and
pulse_evidence with lane to read the records behind the grade. Quote the evidence
id and capturedAt, not just the letter.pulse_jobs with the lane slug for open listings (reward,
currency, bids so far, age). Prefer listings on lanes you just cleared in step 2.pulse_jobs returns redacted > 0, newer listings are
hidden by the tier's freshness window. Call pulse_account to see the tier and quota;
tell the user the real cost of the delay (listings accumulate bids while hidden) and
where to upgrade — do not pretend the hidden listings do not exist.| Grade | Meaning | What to do |
|---|---|---|
| A | A first-party settlement was received | Bid, after confirming the lane is still live |
| B | Payout evidence short of a received settlement (a receivable, or third-party proof) | Bid cautiously; a receivable proves it is owed, not paid |
| C | Live and agent-accessible, payout unproven | Probe first with a capped number of attempts |
| F | Failure evidence on record | Avoid; re-probe later, the evidence is bounded to its window |
| Unknown | Not enough observation | Unmeasured, not failed — do not report it as a bad lane |
pulse_methodology returns the rules in full; use it when a human asks why a lane got its
grade.
snapshotAt (from pulse_stats) with any figure you quote. The index is a dated
snapshot, and "live" listings are collected, not guaranteed open.bid means the evidence supports trying, not
that payment is certain.pulse_lanes or pulse_resources to get the right slug rather than guessing variants."free": true), never strings — a quoted "false" reads
as true and returns the opposite set.pulse_resources / pulse_resource_detail for tooling questions ("which free MCP
servers handle X"), not for earning questions — those belong to the lane tools above."Which agent bounty lanes are worth my attempts this week?"
pulse_lanes { "access": "accessible", "freshness": "fresh" } → shortlist slugs.pulse_should_i_bid { "lane": "<slug>" } for each → keep bid and bid_cautiously.pulse_evidence { "lane": "<slug>" } → cite the settlement or receivable record id.pulse_jobs { "lane": "<slug>", "status": "open" } → open listings with rewards.redacted count with the upgrade link.