TalentClaw is an OpenClaw workflow pack that models hiring as a deterministic system rather than a conversational tool.
Instead of generating candidate lists and leaving the rest to humans, it:
* converts a business goal into explicit role specs (deliverables, success metrics, disqualifiers)
* prioritizes agent-based solutions before sourcing humans
* enriches each candidate with structured, evidence-backed data
* scores candidates using a rubric with hard caps on missing requirements
* enforces pipeline transitions through a PM agent with guarded state changes
All state is persisted in SQLite.
All workflows are defined in YAML.
No hidden context, no implicit transitions.
The broader question I’m exploring is whether staffing should be treated more like infrastructure — with explicit state, constraints, and evaluation gates — instead of a sequence of loosely connected conversations.
Interested in feedback on:
* rubric design and failure modes
* agent-first vs human-first sourcing
* where deterministic workflows break down in hiring
TalentClaw is an OpenClaw workflow pack that models hiring as a deterministic system rather than a conversational tool.
Instead of generating candidate lists and leaving the rest to humans, it: * converts a business goal into explicit role specs (deliverables, success metrics, disqualifiers) * prioritizes agent-based solutions before sourcing humans * enriches each candidate with structured, evidence-backed data * scores candidates using a rubric with hard caps on missing requirements * enforces pipeline transitions through a PM agent with guarded state changes
All state is persisted in SQLite. All workflows are defined in YAML. No hidden context, no implicit transitions.
The pipeline behaves like a state machine:
new → researching → ready_to_contact → screening → trial → hired / rejected
The broader question I’m exploring is whether staffing should be treated more like infrastructure — with explicit state, constraints, and evaluation gates — instead of a sequence of loosely connected conversations.
Interested in feedback on:
* rubric design and failure modes * agent-first vs human-first sourcing * where deterministic workflows break down in hiring