Notebook session ended
This session ended after eight hours.
Sign in again to continue from where you left off.
Notebook gate offline
The sign-in service is not responding right now.
Wait a minute and try again, or read what is already public from the hub.
Foundational walk through the asgard project. Read top to bottom if you're new; no prior context assumed.
More about this project on the Asgard hub.
Use when: orienting in the asgard-codes repo for the first time; deciding which top-level section owns a file you need to find; confirming the file-naming convention before adding a doc
Use when: understanding why there are two agent subsystems in asgard-codes; deciding which agent (Yáo or Jazz) owns a given content flow; locating the shared substrate both agents build on
Use when: explaining what Yáo is (and is not) allowed to do; deciding whether a marketing draft should be routed through Yáo before it ships; onboarding to the two-content-agent split (Yáo social vs jazz educator)
Use when: reconstructing Yáo's methodology in order, not just its outputs; understanding why arc-state is read before the lexicon and anti-patterns; auditing whether a Yáo review actually applied all six principles
Use when: judging a single caption candidate against Devarno's voice; explaining why a grammatically-fine caption still fails scrutiny; building a repeatable rubric instead of ad-hoc taste calls
Use when: assigning a severity to a caption that trips the anti-pattern catalogue; deciding whether a flagged item is a refusal, a warning, or a human call; auditing the catalogue's rule ids and the override recurrence threshold
Use when: structuring a Yáo review so every mandated section is present; understanding why every response appends a decision-log entry; reading the decision_log in arc-state.yaml as training data
Use when: understanding what state Yáo reads before a sequencing decision; reasoning about the staleness gate on feed data; tracing where the canonical arc-state now lives after the abacus move
Use when: tracing how a git commit becomes a queued content suggestion; debugging why a candidate was refused, capped, or rerouted; extending the pipeline with a new gate or stage
Use when: wiring the anthropic-sdk invoker's tool-use loop for YAO generation; adding or auditing a grounding tool the drafting LLM can call; understanding what terminates the bounded tool-use loop
Use when: deciding whether a candidate is refused, softened, or passed; tuning the conformance pre-gate or anti-pattern checks; explaining why scrutiny never calls an LLM
Use when: understanding how unverifiable claims get blocked or capped; reading a target repo's .petrova/contract.yaml claim_index; distinguishing this repo's seam from the external petrova runtime
Use when: tracing the unattended drafting-to-shipped conductor; explaining why ship is dry-run and there is no live stage; debugging why an artifact was held instead of approved
Use when: choosing an LLM invocation backend for the YAO/JAZZ pipeline; wiring a dry-run agent into a test or CI smoke check; adding a new pluggable agent impl behind createAgent
Use when: parsing the return value of any agent-runner invoker; checking why the model stopped before trusting the text; reading the tool-call audit trail after an anthropic-sdk run
Use when: consuming a --json emission from any asgard CLI command; adding a new command that must emit machine-checkable output; validating a persisted results/cli-json envelope
Use when: moving an artifact between lifecycle stages; deciding whether an artifact is allowed to reach approved; auditing why a transition or approval was rejected
Use when: explaining what the jazz (dj-educator) subsystem is and why it exists; deciding whether a ledger record can become a published journal entry; positioning jazz as the second content agent, peer to Yao
Use when: tracing an artefact end-to-end through the jazz subsystem; understanding the translate revise-loop and batch entrypoints; reading a jazz batch report shape
Use when: auditing why an entry failed the jazz gate; adding or reasoning about a deterministic defect check; distinguishing the deterministic layer from the semantic pass
Use when: explaining the autonomy hinge that lets entries auto-approve; debugging why a gate-passing entry was still held; reasoning about the JUDGE_THRESHOLD and fail-closed behaviour
Use when: computing or auditing the slug cost of a jazz entry; reasoning about the gap_penalty > fidelity_rate invariant; understanding what a settlement record freezes and when it goes stale
Use when: onboarding a new org to jazz via a single adapter file; checking the required fields of the entry-output or ledger-artefact contract; explaining the provenance block as the audit receipt
Use when: deriving the display or ship stage for a jazz entry; distinguishing the jazz stage vocabulary from the yao lifecycle stages; debugging why an entry shows held, gated, or stale
Use when: testing an LLM profile against local Ollama models; adding a new validation strategy or CLI command to the legacy layer; distinguishing the profile-testing framework from the content agents
Use when: reasoning about how content ships without a human in the loop; checking whether shipping is live or still fake / dry-run; finding the human step that gates live publishing