partner program
Use Specability Core
on real work.
Specability is not looking for passive signups. The first Core cohort is for teams willing to run the harness inside real agent workflows, show where it breaks, and help turn that evidence into better domain methodology.
the principle
The product learns from work, not surveys.
The CLI is only the delivery surface. The durable asset is the method library: how tasks are framed, delegated, verified, reviewed, recovered, and finished. That library improves fastest when the harness is tested against real codebases, real teams, and real failure modes.
what you get
- one
Specability Core
A free local build of Specability with the starter playbooks/specs needed to run a useful agent workflow. It does not independently upload data to SpecabilityAI; enabled host hooks send bounded orientation to the selected AI provider.
- two
Workflow Review
A focused review of one agent workflow: where the agent loses context, where verification is weak, where delegation is too heavy, and where the process should be encoded.
- three
Domain Method
A bounded method contribution shaped around the workflow you are actually trying to run, kept private while the pattern is still being proven.
who fits
The right team has one painful workflow and enough history to inspect.
This is not a broad beta. The strongest partners are software teams already using coding agents, hitting repeatable workflow failures, and willing to let those failures become product evidence.
good fit
- software teams using Claude Code, Codex CLI, Gemini CLI, or similar agent hosts
- repeatable engineering workflows with real verification needs
- a named owner who can review task frames, handoffs, and outcomes
- willingness to share sanitized feedback about what worked and what failed
not a fit
- teams looking only for prompt tips or tool training
- one-off creative work with no repeatable workflow
- workflows with no observable history, tests, review traces, or outcomes
- organizations that cannot discuss failures, even after sanitization
the loop
A partner is not a user interview. It is a small deployment.
The engagement is intentionally close. We observe the workflow, adjust the method, and decide which patterns are general enough to become part of Specability.
- 01
Frame one workflow
Name the task family, current agent host, expected outcome, verification evidence, and where the workflow currently loses reliability.
- 02
Run Specability Core
Use Specability Core locally. No automatic telemetry is required; feedback is explicit, selected, and reviewable before it leaves your environment.
- 03
Extract the method
Turn repeated problems into task framing rules, playbooks, specs, evidence gates, or recovery behavior.
- 04
Productize what repeats
Promote only generalized patterns into the relevant domain overlay. Customer-specific context stays out of public docs and released methodology unless explicitly cleared.
the boundary
Specability Core is free. The full method library is not public.
The public distribution repository exists for checksummed builds, changelog, install guidance, and feedback. The full source, internal experiments, unreleased domain methodology, and complete method library remain private until explicitly released.
Bring one workflow worth fixing.