The Great Libraryof SISO

Library / Agents / Promotion

Promotion

Evidenced agent mechanisms moving through explicit ownership, portability, verification, release, and Library gates.

P—01

Campaign / agent-capability-promotion

SISO agent capability promotion campaign

Continuously turn evidenced local agent mechanisms into owned, sanitized, verified, released, Library-indexed, and stack-pinned capabilities without importing private state.

partial · observed 2026-08-01
  1. 01unverified
  2. 02read
  3. 03candidate
  4. 04owner assigned
  5. 05extracted
  6. 06verified
  7. 07released
  8. 08library indexed
  9. 09stack pinned
  10. 10retired

Campaign / foundry-agency-intelligence

Foundry Agency intelligence promotion inventory

Turn recovered source intelligence into evidence-backed Agency OS capability decisions and promote verified results to their owning Works without copying private corpora or source-owned applications into the Library.

partial · observed 2026-08-03
  1. 01unverified
  2. 02read
  3. 03candidate
  4. 04owner assigned
  5. 05extracted
  6. 06verified
  7. 07released
  8. 08library indexed
  9. 09stack pinned
  10. 10retired

Campaign totals

25 candidates

Across 2 accepted inventories.

Stage

candidate11

owner assigned8

read5

released1

Priority

critical13

high10

medium2

Target

Foundry external data plane behind a read-only Source Oracle contract1

Optional SISO Agent Hooks efficiency profile1

Runtime evaluation packages and Agent Stack release proof1

SISO Agent Brain or Session Intelligence optional lifecycle integration1

SISO Agent Hooks1

SISO Agent Hooks completion profile and Project OS reconciliation probe1

SISO Agent Hooks evidence profile1

SISO Agent Hooks optional policy pack1

SISO Agent Integrations initially; independent Atlas Work only after activity proof1

SISO Agent Playbook1

SISO Agent Runtime control core1

SISO Agent Runtime incubation or an independently evidenced Harness Lab Work1

SISO Agent Runtime operator surfaces1

SISO Agent Runtime TUI packages and optional host adapters1

SISO Agent Zero optional voice adapter1

SISO Foundry1

SISO Foundry evidence intake; future SISO Agency OS module-portfolio owner is unassigned1

SISO Foundry evidence intake; future SISO Agency OS product owner is unassigned1

SISO Foundry verification and outcome evidence1

SISO Foundry; future Agency owner remains unassigned1

SISO Project OS drift checks and Great Library operating docs1

SISO Project OS standards and verifier package1

SISO Skills1

SISO Skills next reviewed release1

SISO Skills, SISO Agent Playbook, and SISO Agent Stack Distribution1

Promotion units

Candidate gates

Evidence-led status, not release status.

owner assignedcritical priorityhigh confidencerequires sanitization

Hooks registry and promotion governance

Classification
shared package
Disposition
extract
Source scope
agent-os-workspace
Target
SISO Agent Hooks
Evidence owner Works
Same as product owner
Next gate
Reconcile the legacy registry with the public manifest and add machine-verifiable maturity, portability, test, and latency fields.
Blockers
  • The historical staging lifecycle was documented but never implemented.
  • Several registry entries describe specifications or machine-bound implementations rather than verified portable hooks.
Rationale
The public Hooks Work distributes a curated baseline, while the older hub contains useful registry, audit, template, health, and promotion concepts that remain unproductized.
Evidence · 1 record
  • source-review:agent-capability-promotion:hooks-governance:2026-08-01
    source_review · 2026-08-01
candidatecritical priorityhigh confidencerequires sanitization

Manifest-driven deterministic policy guards

Classification
shared package
Disposition
extract
Source scope
agent-host-configuration
Target
SISO Agent Hooks optional policy pack
Evidence owner Works
Same as product owner
Next gate
Separate generic destructive-action, authority, and repository-safety rules from personal routing policy and add false-positive fixtures.
Blockers
  • The source combines portable safeguards with operator-specific model and path policy.
Rationale
The local guard enforces a broader collection of prevention rules than the released safety hook, but its policy and host adapter are not yet separated.
Evidence · 1 record
  • source-review:agent-capability-promotion:deterministic-guards:2026-08-01
    source_review · 2026-08-01
candidatecritical priorityhigh confidencerequires sanitization

Dispatch evidence contract hook

Classification
shared package
Disposition
extract
Source scope
agent-host-configuration
Target
SISO Agent Hooks evidence profile
Evidence owner Works
Same as product owner
Next gate
Define a provider-neutral worker result schema and test numeric, existence, directional, and classification claims against required evidence.
Blockers
  • The current parser is coupled to one host transcript format.
Rationale
The Playbook describes evidence-backed worker returns, but stack recipients do not receive the local mechanism that checks those claims.
Evidence · 1 record
  • source-review:agent-capability-promotion:evidence-contract:2026-08-01
    source_review · 2026-08-01
candidatecritical priorityhigh confidencerequires sanitization

Closeout, earned-ask, and salvage gates

Classification
shared package
Disposition
extract
Source scope
agent-host-configuration
Target
SISO Agent Hooks completion profile and Project OS reconciliation probe
Evidence owner Works
Same as product owner
Next gate
Extract a conservative policy core, define host adapters, and build precision and recall fixtures for completion and operator-attention decisions.
Blockers
  • Transcript parsing and worktree discovery contain host and fleet assumptions.
  • Aggressive stop gates require false-positive measurement before default installation.
Rationale
The local completion system covers unfinished work, unsupported done claims, unnecessary hand-backs, and stranded peer work beyond the released mutation-verification hook.
Evidence · 1 record
  • source-review:agent-capability-promotion:closeout-attention:2026-08-01
    source_review · 2026-08-01
candidatehigh priorityhigh confidencerequires sanitization

Batching and delegation efficiency hooks

Classification
shared package
Disposition
extract
Source scope
agent-host-configuration
Target
Optional SISO Agent Hooks efficiency profile
Evidence owner Works
Same as product owner
Next gate
Make thresholds configurable, reduce counters to publication-safe aggregates, and discover structured-search capability dynamically.
Blockers
  • Current thresholds and indexed-project knowledge reflect one operator environment.
Rationale
These hooks mechanically reinforce batching, bounded delegation, and structured code navigation that currently exist mostly as written operating guidance.
Evidence · 1 record
  • source-review:agent-capability-promotion:efficiency-hooks:2026-08-01
    source_review · 2026-08-01
owner assignedcritical priorityhigh confidenceunknown

Warehouse-to-Hub skill reconciliation

Classification
shared package
Disposition
compare
Source scope
agent-base-warehouse
Target
SISO Skills
Evidence owner Works
Same as product owner
Next gate
Read and classify every catalog delta, promote only current portable mechanisms, repair metadata, and record explicit retirement or Playbook ownership for the rest.
Blockers
  • Name-level comparison is only a nomination signal.
  • Sampled entries include retired, malformed, nested, private, and scenario-sized material.
Rationale
The warehouse contains substantially more skill documents than the released catalog, and a filtered audit found many names absent from both Skills and Playbook without proving that every delta should ship.
Evidence · 1 record
  • source-review:agent-capability-promotion:skills-delta:2026-08-01
    source_review · 2026-08-01
candidatehigh priorityhigh confidencerequires sanitization

Priority portable skill batch

Classification
shared package
Disposition
extract
Source scope
agent-base-warehouse
Target
SISO Skills next reviewed release
Evidence owner Works
Same as product owner
Next gate
Repair discovery metadata, preserve attribution, add trigger and negative tests, and verify bundled scripts with content-neutral fixtures.
Blockers
  • Two candidates have incomplete or malformed discovery metadata.
  • The video skill requires fixture-based script verification before publication.
Rationale
Direct review found a small high-value batch spanning fact verification, scope control, falsifiable investigation, volatile-state reminders, and video analysis.
Evidence · 1 record
  • source-review:agent-capability-promotion:priority-skills:2026-08-01
    source_review · 2026-08-01
readhigh priorityhigh confidencehost specific

Pi Harness Lab

Classification
experiment
Disposition
compare
Source scope
agent-base-warehouse
Target
SISO Agent Runtime incubation or an independently evidenced Harness Lab Work
Evidence owner Works
Same as product owner
Next gate
Run parity analysis against released Runtime packages and identify code, tests, and research that remain uniquely useful.
Blockers
  • The lab combines provider-specific experiments, generated patches, live verification, and functionality partly promoted elsewhere.
  • Independent adoption has not been established.
Rationale
The lab has real packages and extensive local and live smoke surfaces for routing, lazy skills, child agents, councils, workflows, dashboards, lifecycle restore, and token budgets.
Evidence · 1 record
  • source-review:agent-capability-promotion:pi-harness-lab:2026-08-01
    source_review · 2026-08-01
owner assignedcritical priorityhigh confidenceready

Provider-neutral child-task lifecycle

Classification
shared package
Disposition
extract
Source scope
agent-base-warehouse
Target
SISO Agent Runtime control core
Evidence owner Works
Same as product owner
Next gate
Implement the typed state machine and an exactly-once notification regression fixture in Runtime control core.
Blockers
  • The mechanism currently exists as a specification rather than released Runtime code.
Rationale
The specification separates liveness, terminal result, output, and usage while requiring terminal-before-embellishment ordering and exactly-once parent notification.
Evidence · 1 record
  • source-review:agent-capability-promotion:child-task-lifecycle:2026-08-01
    source_review · 2026-08-01
candidatehigh priorityhigh confidencehost specific

SISO TUI application and adapter parity

Classification
shared package
Disposition
compare
Source scope
agent-base-warehouse
Target
SISO Agent Runtime TUI packages and optional host adapters
Evidence owner Works
Same as product owner
Next gate
Extract renderer-neutral components and status adapters, then measure the remaining provider-specific application boundary.
Blockers
  • The released Runtime intentionally contains contracts rather than raw application and renderer parity.
  • Pi-native patches require an optional adapter boundary.
Rationale
The active terminal application contains renderer contracts, layouts, runtime and status adapters, and interaction code beyond the currently released TUI contracts package.
Evidence · 1 record
  • source-review:agent-capability-promotion:tui-parity:2026-08-01
    source_review · 2026-08-01
owner assignedcritical priorityhigh confidencerequires sanitization

Agent-legible repository standards

Classification
shared package
Disposition
extract
Source scope
workspace-standards
Target
SISO Project OS standards and verifier package
Evidence owner Works
Same as product owner
Next gate
Reconcile overlapping rules, separate portable templates from workspace examples, and add cold-agent orientation and structure conformance checks.
Blockers
  • The source combines general standards with historical workspace-specific examples and personal paths.
Rationale
The sources define a useful cold-agent orientation test, boot placement, depth and naming rules, universal agent primitives, ADR governance, structural ratchets, and handoff contracts.
Evidence · 1 record
  • source-review:agent-capability-promotion:agent-legibility:2026-08-01
    source_review · 2026-08-01
owner assignedhigh priorityhigh confidenceready

Manager–Worker–Validator artifact improvement playbook

Classification
shared package
Disposition
extract
Source scope
agent-base-warehouse
Target
SISO Agent Playbook
Evidence owner Works
Same as product owner
Next gate
Package role prompts, file-handoff schemas, exactly-one-change enforcement, retry bounds, and an unintended-change fixture.
Blockers
  • The current mechanism is warehouse documentation rather than an installable verified scenario.
Rationale
The triad offers a disciplined reusable loop: analysis-only manager, exactly-one-change worker, and binary validator with unintended-change auto-failure.
Evidence · 1 record
  • source-review:agent-capability-promotion:manager-worker-validator:2026-08-01
    source_review · 2026-08-01
owner assignedcritical priorityhigh confidencerequires sanitization

Agent Stack benchmark and proof harness

Classification
shared package
Disposition
extract
Source scope
agent-base-warehouse
Target
Runtime evaluation packages and Agent Stack release proof
Evidence owner Works
Same as product owner
Next gate
Sanitize cases, define a stable scorecard schema, and make clean-install, routing, lifecycle, fault, privacy, and recovery checks release evidence for the whole stack.
Blockers
  • Some existing cases and reports encode retired hosts, private routing, or historical environment facts.
Rationale
The harness already models task success, efficiency, code navigation, delegation, durability, verification, coordination, skills, safety, memory, fault injection, and soak tiers.
Evidence · 1 record
  • source-review:agent-capability-promotion:benchmark-harness:2026-08-01
    source_review · 2026-08-01
owner assignedhigh priorityhigh confidencerequires sanitization

Agent Zero replay-only voice contracts

Classification
experiment
Disposition
extract
Source scope
agent-base-warehouse
Target
SISO Agent Zero optional voice adapter
Evidence owner Works
Same as product owner
Next gate
Promote typed intent, binding, receipt, handoff, risk, idempotency, and replay fixtures while retaining dry-run-only transport.
Blockers
  • Live dispatch remains intentionally gated.
  • Current target discovery and lane admission include operator environment assumptions.
Rationale
The extension contains a safer typed and fail-closed interface that the released Agent Zero Work currently records only as deferred capability.
Evidence · 1 record
  • source-review:agent-capability-promotion:agent-zero-voice:2026-08-01
    source_review · 2026-08-01
readmedium prioritymedium confidencerequires sanitization

Atlas read-only architectural auditor

Classification
experiment
Disposition
compare
Source scope
agent-base-warehouse
Target
SISO Agent Integrations initially; independent Atlas Work only after activity proof
Evidence owner Works
Same as product owner
Next gate
Prove a live consumer and activity history, then sanitize the rubric and dependency surface before incubation.
Blockers
  • Operational usage remains unverified.
  • The rubric includes personal and business-specific criteria.
Rationale
Atlas defines a distinct graph-assisted, read-only investigation loop with three-critic triangulation, deduplication, and ticket emission only after agreement.
Evidence · 1 record
  • source-review:agent-capability-promotion:atlas:2026-08-01
    source_review · 2026-08-01
candidatehigh priorityhigh confidencehost specific

Cross-host agent operator HUD

Classification
shared package
Disposition
extract
Source scope
agent-host-configuration
Target
SISO Agent Runtime operator surfaces
Evidence owner Works
Same as product owner
Next gate
Define a provider-neutral status input contract and capability-detected Claude and Codex render adapters.
Blockers
  • The existing statusline consumes one host's transcript and rate-limit surfaces.
Rationale
The dependency-free statusline provides model, context pressure, token flow, reset, rate-limit, and session-cost feedback missing from the portable stack.
Evidence · 1 record
  • source-review:agent-capability-promotion:operator-hud:2026-08-01
    source_review · 2026-08-01
readmedium prioritymedium confidenceprivate overlay

Memory lifecycle hook adapter

Classification
experiment
Disposition
compare
Source scope
agent-os-workspace
Target
SISO Agent Brain or Session Intelligence optional lifecycle integration
Evidence owner Works
Same as product owner
Next gate
Model lifecycle events without personal memory payloads and run a privacy threat review before any extraction.
Blockers
  • The current capability is coupled to plugin-local memory and host lifecycle events.
  • No personal memory content may enter the public source or fixtures.
Rationale
The lifecycle configuration suggests reusable SessionStart, PostToolUse, and Stop integration, but only the event contract is eligible for public review.
Evidence · 1 record
  • source-review:agent-capability-promotion:memory-lifecycle:2026-08-01
    source_review · 2026-08-01
owner assignedhigh priorityhigh confidenceready

Generated workspace ownership and domain canon

Classification
shared package
Disposition
extract
Source scope
workspace-standards
Target
SISO Project OS drift checks and Great Library operating docs
Evidence owner Works
Same as product owner
Next gate
Generate human summaries from the domain manifest and fail verification when declared repository counts or ownership boundaries drift.
Blockers
  • Current root documents are hand-maintained and disagree with the expanded ten-repository Agents domain.
Rationale
The live domain manifest records the current estate, while root orientation and ownership documents still contain older repository counts and incomplete boundaries.
Evidence · 1 record
  • source-review:agent-capability-promotion:workspace-canon:2026-08-01
    source_review · 2026-08-01
releasedcritical priorityhigh confidenceready

Foundry Agency intelligence package

Classification
shared package
Disposition
keep
Source scope
foundry-public-work
Target
SISO Foundry
Evidence owner Works
Same as product owner
Next gate
Select the accepted Foundry 0.4.0 Release in a successor Library Snapshot; keep Agency coverage receipts and ownership boundaries current as Foundry publishes later revisions.
Blockers

None recorded.

Rationale
The package makes the recovered Agency mission, metadata counts, application map, caveats, and evidence backlog durable inside the Work that owns source intelligence rather than burying them in an agent transcript or duplicating them in the Library.
Evidence · 2 records
  • source-review:foundry-agency-intelligence-package:2026-08-02
    source_review · 2026-08-03
  • release-receipt:foundry-0.4.0
    release_receipt · 2026-08-02
candidatecritical priorityhigh confidenceready

Foundry Agency coverage inventory v0.5.1

Classification
shared package
Disposition
keep
Source scope
foundry-public-work
Target
SISO Foundry; future Agency owner remains unassigned
Product owner Works
No product owner assignedOwner registration is the next gate.
Next gate
Phase 2: direct-source dossiers and integration proofs for the ten specified Agency integrations, followed by an owner decision; do not infer source-code reuse rights or runtime adoption authorization from this coverage receipt.
Blockers
  • Coverage is published metadata only; no Agency product Work owns these candidates and no runtime adoption authorization is granted.
Rationale
Foundry v0.5.1 publishes a coverage inventory, 20-row decision matrix, and verified job-level pillar mapping over 628 projects. It is analyzed coverage, not a reusable source bundle: 89 projects are source-read/reusable and 10 confirmed; ten integration proofs are specified and one named script is runtime-proven.
Evidence · 1 record
  • source-receipt:foundry-agency-coverage:v0.5.1
    integration_check · 2026-08-03
candidatecritical priorityhigh confidencerequires sanitization

Source Oracle, adoption truth, and dependency evidence

Classification
data or archive
Disposition
move to data store
Source scope
foundry-operational-evidence
Target
Foundry external data plane behind a read-only Source Oracle contract
Evidence owner Works
Same as product owner
Next gate
Version a typed read-only API and create schema, content-hash, rights, privacy, generator, and durable-copy receipts for an external data snapshot.
Blockers
  • The corpus payload is not published.
  • 879 adoption records remain unresolved.
  • Current external assets do not yet carry the complete release receipt set.
Rationale
Repository identity, adoption decisions, capability dependencies, contract cards, and verification receipts are the compounding source-intelligence asset. Agents should query them through a stable contract rather than receive direct database authority.
Evidence · 2 records
  • source-review:foundry-operational-metadata:2026-08-02
    source_review · 2026-08-03
  • sha256:cfb73b50405fcd71383440cfbcee3d4b896c549a206d3840194de26d50fb843c
    source_review · 2026-08-03
candidatecritical priorityhigh confidenceready

Behavior and composition proof harness

Classification
experiment
Disposition
keep
Source scope
foundry-public-work
Target
SISO Foundry verification and outcome evidence
Evidence owner Works
Same as product owner
Next gate
Publish reproducible fixtures for the WILD experiments and require every Agency module verdict to carry behavioral, integration, cost, and outcome evidence.
Blockers
  • Current composition results support narrow verification leverage, not automatic application assembly.
Rationale
Foundry creates leverage only when source selection is tied to falsifiable behavior and integration outcomes. The existing evidence supports contract cards as steering and validation aids while rejecting stronger Lego-style composition claims.
Evidence · 1 record
  • source-review:foundry-composition-receipts:2026-08-02
    integration_check · 2026-08-03
readcritical prioritymedium confidenceunknown

SISO-owned business control-plane gaps

Classification
candidate work
Disposition
distribute to owners
Source scope
foundry-operational-evidence
Target
SISO Foundry evidence intake; future SISO Agency OS product owner is unassigned
Product owner Works
No product owner assignedOwner registration is the next gate.
Next gate
Register or identify the owning Agency control-plane Work, specify the six gaps as product contracts, and rerun source discovery against each contract before authorizing a build.
Blockers
  • No accepted Agency OS or business control-plane Work currently exists in the Library.
  • A source gap is a build hypothesis, not proof that suitable source does not exist.
Rationale
Foundry owns the candidate evidence only, not the future product. Usage metering, pricing and entitlements, multi-tenancy, append-only business audit, reliable webhook delivery, and OAuth authority recur across the source portfolio without one proven drop-in owner. They are likely SISO leverage only after contract definition, renewed evidence, and Agency owner registration.
Evidence · 1 record
  • source-review:foundry-agency-control-plane-gaps:2026-08-02
    source_review · 2026-08-03
candidatehigh priorityhigh confidencerequires sanitization

Agent-system pattern and prompt bank

Classification
shared package
Disposition
distribute to owners
Source scope
foundry-operational-evidence
Target
SISO Skills, SISO Agent Playbook, and SISO Agent Stack Distribution
Evidence owner Works
Same as product owner
Next gate
Review every promoted pattern against its direct source, attribution, rights, behavior, overlap, and owner contract before extraction or Playbook integration.
Blockers
  • The 62 patterns and 26 prompt records are research evidence, not independently licensed release artifacts.
  • Owner-specific tests and packaging do not yet exist.
Rationale
Foundry has already mined useful agent patterns and prompt evidence. The value becomes durable only when the appropriate owning Work turns selected mechanisms into attributed, verified, installable releases.
Evidence · 1 record
  • source-review:foundry-agent-pattern-bank:2026-08-02
    source_review · 2026-08-03
readhigh prioritymedium confidenceunknown

Agency source-module candidate portfolio

Classification
external upstream
Disposition
compare
Source scope
agency-source-portfolio
Target
SISO Foundry evidence intake; future SISO Agency OS module-portfolio owner is unassigned
Product owner Works
No product owner assignedOwner registration is the next gate.
Next gate
Create direct-source dossiers and integration proofs for each capability cluster, then assign accepted modules to an Agency portfolio Work instead of treating a star-ranked list as architecture.
Blockers
  • License, security, maintenance, tenancy, API coverage, deployment burden, and current source revisions require direct review.
  • No accepted Agency OS portfolio Work currently owns these candidates.
Rationale
Foundry owns the candidate evidence only, not the future product portfolio. Teable, AFFiNE, Open Notebook, Plane, Documenso, Chatwoot, Cal.com, Activepieces, Listmonk, Postiz, Formbricks, Metabase, PostHog, Superset, Meilisearch, Qdrant, RAGFlow, Coolify, Airbyte, dlt, Meltano, Langflow, Coze, and Flowgram are operator-nominated starting ideas, but each requires a current evidence-backed adoption verdict and an Agency owner.
Evidence · 1 record
  • source-review:foundry-agency-module-portfolio:2026-08-02
    source_review · 2026-08-03