Talent Summoner
Server Details
AI candidate sourcing: find, review and shortlist talent worldwide from your AI assistant.
- Status
- Healthy
- Uptime
- 100.0% over the last 24 hours
- Last Tested
- Transport
- Streamable HTTP · MCP 2025-11-25
- URL
- Repository
- talent-summoner/agent-skills
- GitHub Stars
- 0
- ChatGPT
- –
- Claude
- –
TDQS
Scored across 10 tools
The tools mostly map to distinct steps in the sourcing lifecycle: start/continue/read/stop, candidate review, contact lookup, outreach preview/send, sharing, and role unlock. There is mild overlap in the outreach area, since sourcing_continue can request outreach drafts while sourcing_outreach previews them and sourcing_read exposes shared outreach templates, but the descriptions make the boundaries workable.
All tools use a consistent snake_case naming scheme with the same sourcing_ prefix. Compound action names like sourcing_outreach_send fit the same pattern and remain predictable.
Ten tools is well scoped for a candidate-sourcing workflow, with each tool covering a real lifecycle stage. Although sourcing_outreach and sourcing_share are multi-action tools, the overall count is not bloated.
The surface covers session creation, continuation, reading/listing, stopping, candidate review, contact enrichment, outreach preview/send, public sharing, and paid-role unlock. Minor lifecycle gaps remain, such as no explicit delete/archive operation for a sourcing session.
Available Tools
10 toolssourcing_contactLook up candidate work emailADestructiveInspect
Start a work-email lookup for one candidate in an owned sourcing session through an external contact-enrichment provider. Returns emailLookupStatus; the lookup finishes in the background and its result appears on the candidate in sourcing_read. Contact details stay subject to the session's masking.
Inputs: sessionId (required), candidateId (required).
sourcing_continueSend message to sourcing sessionADestructiveInspect
Send one user message to an owned sourcing session, such as an ICP edit or approval, a request to search or find more candidates, a refinement, or an outreach-draft request. Returns sessionId, activity and appUrl once the turn is accepted; the work continues in the background and its results appear in sourcing_read. A second sourcing_continue while a turn is running replaces that turn. Searches never assign one of the user's paid Roles: evaluation that needs a Role stops, sourcing_read reports billing.paymentRequired, and only sourcing_unlock_role assigns one. Outreach drafts cover shortlisted candidates only; sourcing_read returns their shared template, while sourcing_outreach reviews personalised messages and sourcing_outreach_send confirms sending.
Inputs: userText (required), sessionId (required).
sourcing_outreachReview outreachADestructiveInspect
Preview personalised email or LinkedIn outreach for an owned session, read one confirmed batch status, or resolve shortlisted candidates' LinkedIn reachability. Preview returns the sender, deterministic full-batch recipient/message pages, draftId and reviewFingerprint; recipientOffset/recipientLimit and bodyOffset page the result until complete is true, and a changed draftId or reviewFingerprint invalidates the prior review. Status requires the batchId returned by sourcing_outreach_send. Email recovery remains available within 30 minutes of confirming. LinkedIn confirmations remain queued for up to 14 days and send in the background; pending LinkedIn rows resume without an active turn. Resolve may call the connected LinkedIn provider and update cached reachability. Returns safe recipient aliases, readiness, delivery states and connection hand-off details. Failed recipients carry errorCode, a closed plain reason (e.g. message_too_long); provider_rejected means no reason mapped. A result with simulated: true comes from a review-sandbox account: nothing was delivered or looked up.
Inputs: action (required), channel (required), sessionId (required), batchId, +3 more.
sourcing_outreach_sendConfirm and send outreachADestructiveInspect
Confirm outreach through the connected email or LinkedIn account after the user has reviewed every page of the preview and approved the exact sender, recipients and messages. First confirmation requires sessionId, channel, draftId and reviewFingerprint from sourcing_outreach; LinkedIn also requires linkedinRiskAcknowledged true. Email starts in the current assistant turn and recovery accepts the confirmed batchId within 30 minutes. LinkedIn returns queued and sends in the background for up to 14 days; it does not need an assistant turn to continue. A recipient listed in priorContacts of the preview (someone in the workspace already contacted them) is refused with 409 prior_contact_unconfirmed unless the user approved messaging them again and their candidateId is in confirmedPriorContactCandidateIds. Returns batchId, recipient count and queued, sending, resumed, replayed or confirmed_not_started status. Confirmation is irreversible and may contact external recipients. A result with simulated: true comes from a review-sandbox account: nothing was delivered or looked up.
Inputs: channel (required), sessionId (required), batchId, draftId, +3 more.
sourcing_readRead sourcing sessionARead-onlyIdempotentInspect
Read owned session; omit sessionId to list. Returns full detail: question, appUrl, candidates and transcript. Under-bar rows are listed, marked hiddenReason "below_minimum"; current staff are not. Pages default to 10 candidates and 10 transcript messages. Page with candidateOffset and candidateLimit (max 50), transcriptOffset and transcriptLimit (max 20); transcriptOffset counts back from newest. candidatePage (with missingCandidateIds) and transcriptPage return offset, limit, total and hasMore. view "status" is lightweight. Filter by candidateIds (max 50), selection or filterStatus; sort by candidateSort "discovery" or "fit". outreachDraft is the shared template; sourcing_outreach has personalised messages and delivery status. Contacts masked until paid. billing.paymentRequired is true for a current block, null when unverifiable, else false; billing.remainingFreshEvaluations is paid evaluations left. If textLimits is present, detail was reduced to fit the client's result limit; for full detail, read fewer candidates (candidateIds or a smaller candidateLimit) with transcriptLimit: 1. textLimits.icp caps ICP at 10,000 characters; paging does not change it; full ICP is at appUrl.
Inputs: view, limit, cursor, selection, +8 more.
sourcing_reviewReview candidateADestructiveIdempotentInspect
Save the user's review decision (liked, shortlisted or skipped) and/or hiring-funnel stage for one candidate in an owned sourcing session, with optional reasons and note. The stage is contacted, replied, interviewed, offer, hired, retained, or a custom stage: any non-blank text of at most 40 characters; empty, blank and over-long values are rejected. outcomeStage null clears it. Pass selection, outcomeStage, or both; reasons and note require selection. The two are independent: an omitted selection keeps the saved decision, and an omitted outcomeStage keeps the saved stage. Replaces the candidate's previous decision when selection is sent; an omitted reasons or note keeps its saved value, and an empty list or string clears it. candidateId is a candidate id from sourcing_read. Returns the saved decision and/or stage.
Inputs: sessionId (required), candidateId (required), note, reasons, +2 more.
sourcing_startStart sourcing searchADestructiveInspect
Create a sourcing session from a hiring brief and start its first turn. startMode "review_criteria" (default) drafts an ideal candidate profile (ICP) for review; "find_candidates" searches for candidates immediately. Returns sessionId, activity and appUrl as soon as the turn is accepted; the work continues in the background and its results appear in sourcing_read. Searches never assign one of the user's paid Roles: evaluation that needs a Role stops, sourcing_read reports billing.paymentRequired, and only sourcing_unlock_role assigns one.
Inputs: prompt (required), startMode.
sourcing_stopStop sourcing turnADestructiveIdempotentInspect
Cancel the active turn of an owned sourcing session. Returns cancelled: true when a running turn was stopped, false when nothing was running. Candidates and results produced before the stop stay saved.
Inputs: sessionId (required).
sourcing_unlock_roleUnlock paid Role for sessionADestructiveIdempotentInspect
Assign one of the user's available paid Roles to an owned sourcing session, unlocking paid access for it. This spends the Role and cannot be undone; a session that is already paid keeps its Role and no other is spent. Returns sessionId. Fails with status 402 when no Role is available and 403 when the user's email is not verified.
Inputs: sessionId (required).
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Related MCP Connectors
Find your next great hire. Review candidates and manage hiring through your AI assistant.
Hire tested remote staff from anywhere: search, shortlist and send intros from your AI app.
Score resumes, screen candidates by phone, and check references from your AI assistant.
Hire from your AI chat: search 210k+ vetted startup candidates, review matches, request intros.
Related MCP Servers
AlicenseNot gradedqualityCmaintenanceConnects AI assistants to a pool of remote professionals, enabling candidate searches, anonymized profile retrieval, and introduction requests.1MIT- FlicenseNot gradedqualityCmaintenanceGenerate tailored job descriptions, hire senior developers and pull technical interview questions without leaving the conversation.48-
- FlicenseNot gradedqualityAmaintenanceEnables AI assistants to automate recruiter-side candidate sourcing, resume screenshot/OCR viewing, multi-keyword search and deduplication, scoring and filtering, greeting, and Markdown report export.-
- FlicenseNot gradedqualityAmaintenanceEnables AI assistants to automate recruiter-side hiring workflows by batch-searching candidates, capturing resume screenshots, extracting share links, filtering and scoring profiles, sending greetings, and exporting Markdown reports.-
Glama MCP Gateway
Add one secure layer between your agents and this server.