Human-agent operations
Human-agent operations: orchestration, runtime boundaries, memory, evaluation, security, and cost accounting. Token economics is a named sub-category.
Shared lane
This lane is carried by both hubs of this survey.
Entries are repositories — libraries, benchmarks, specifications, whatever a ledger admitted as evidence. Some of it is polished; most of it is parts.
Tasks in this category
- Design agent organizations 0 of 86 entries
- Run a solo-operator agent firm 0 of 86 entries
- Orchestrate multi-agent workflows 0 of 86 entries
- Agent runtime & boundaries 0 of 86 entries
- Manage agent memory & context 0 of 86 entries
- Agent identity & interoperability 0 of 86 entries
- Evaluate agents & skills 0 of 86 entries
- Token economics & AI cost 10 of 86 entries
- Secure the agent supply chain 0 of 86 entries
- Package & port agent skills 0 of 86 entries
- Operator & research consoles 0 of 86 entries
- Track tool & repo change 0 of 86 entries
- Allocate tasks across humans & agents Planned 0 of 86 entries
- Score orchestration Planned 0 of 86 entries
- MCP/tool interoperability Planned 0 of 86 entries
Token economics & AI cost
Campaigns placed on this task
Multiple runs of this task exist; each is its own sealed record — no supersession is implied.
-
2026-08-01__token-usage-observability-accounting
run type (the campaign's own designation) high-recall-map 10 of 86 entries ingested
- Ledger rows
- 41
- Screening rows
- 113
- Repos in ledger
- 14
-
2026-08-02__token-efficiency-routing-and-optimization
run type (the campaign's own designation) high-recall-map Not yet ingested
- Ledger rows
- 55
- Screening rows
- 160
- Repos in ledger
- 25
Entries — 10 of 86 on this hub
-
Human-agent operations Live
AgentOps-AI/agentops
Token economics & AI cost
Editorial draft Session-level agent observability in Python: token and cost tracking, tool calls, replay. Cited twice in the token-economics ledger.
Forks 612 Language Python Stars 5761
Snapshot · retrieved UTC
-
Human-agent operations Live
Arize-ai/openinference
Token economics & AI cost
Editorial draft OpenTelemetry-compatible tracing conventions and instrumentations for model applications, with masking support.
Forks 288 Language Python Stars 1138
Snapshot · retrieved UTC
-
Human-agent operations Live
BerriAI/litellm
Token economics & AI cost
Editorial draft A multi-provider model gateway; the ledger cites its spend-tracking dimensions and its public issue history as cost-accounting evidence.
Forks 10443 Language Python Stars 55954
Snapshot · retrieved UTC
-
Human-agent operations Live
crewAIInc/crewAI
Token economics & AI cost
Editorial draft A multi-agent framework in Python; two survey ledgers cite it — for usage-accounting fields and for role-and-process organization design.
Forks 8107 Language Python Stars 56858
Snapshot · retrieved UTC
-
Human-agent operations Live
future-agi/traceAI
Token economics & AI cost
Editorial draft OpenTelemetry tracing that instruments model calls, token counts and agent decisions.
Forks 38 Language Python Stars 210
Snapshot · retrieved UTC
-
Human-agent operations Live
lmnr-ai/lmnr
Token economics & AI cost
Editorial draft An OpenTelemetry-native tracing and evaluation platform for agents, in TypeScript.
Forks 220 Language TypeScript Stars 3155
Snapshot · retrieved UTC
-
Human-agent operations Live
open-telemetry/semantic-conventions
Token economics & AI cost
Editorial draft The OpenTelemetry semantic-conventions registry; kept as the source the GenAI conventions moved out of, with the successor listed separately.
Forks 377 Language Jinja Stars 627
Snapshot · retrieved UTC
-
Human-agent operations Live
open-telemetry/semantic-conventions-genai
Token economics & AI cost
Editorial draft Current home of the GenAI semantic conventions — the vocabulary that token and cost telemetry is standardizing on.
Forks 75 Language Python Stars 234
Snapshot · retrieved UTC
-
Human-agent operations Live
openlit/openlit
Token economics & AI cost
Editorial draft OpenTelemetry-native tracing and metrics with a model cost registry, in TypeScript.
Forks 350 Language TypeScript Stars 2676
Snapshot · retrieved UTC
-
Human-agent operations Live
traceloop/openllmetry
Token economics & AI cost
Editorial draft Apache-2.0 OpenTelemetry instrumentations spanning model and agent frameworks.
Forks 1047 Language Python Stars 7368
Snapshot · retrieved UTC
Tasks with no entry today
14 of 15 tasks in this category carry no entry today. Each is listed with its state: ingested with nothing admitted, on the record and not yet ingested, or planned and not yet run.
- Design agent organizations Not yet ingested pre-standard 2 campaigns on the record: one predates the current standard and carries flags, not counts; the other has not been brought into the registry.
- Run a solo-operator agent firm Not yet ingested pre-standard 2 campaigns on the record: one predates the current standard and carries flags, not counts; the other has not been brought into the registry.
- Orchestrate multi-agent workflows Not yet ingested One campaign on the record; its ledger rows have not been brought into the registry.
- Agent runtime & boundaries Not yet ingested 2 campaigns on the record; their ledger rows have not been brought into the registry.
- Manage agent memory & context Not yet ingested pre-standard This campaign predates the current standard; it carries flags, not counts.
- Agent identity & interoperability Not yet ingested pre-standard 2 campaigns on the record: one predates the current standard and carries flags, not counts; the other has not been brought into the registry.
- Evaluate agents & skills Not yet ingested pre-standard 2 campaigns on the record: one predates the current standard and carries flags, not counts; the other has not been brought into the registry.
- Secure the agent supply chain Not yet ingested 2 campaigns on the record; their ledger rows have not been brought into the registry.
- Package & port agent skills Not yet ingested One campaign on the record; its ledger rows have not been brought into the registry.
- Operator & research consoles Not yet ingested pre-standard 2 campaigns on the record: one predates the current standard and carries flags, not counts; the other has not been brought into the registry.
- Track tool & repo change Not yet ingested One campaign on the record; its ledger rows have not been brought into the registry.
- Allocate tasks across humans & agents Planned No campaign has run yet. Grounded in crosswalk row E01 (internal crosswalk id).
- Score orchestration Planned No campaign has run yet. A grounding row is recorded, but its only citation is to material this site does not publish.
- MCP/tool interoperability Planned No campaign has run yet. Grounded in crosswalk row N07 (internal crosswalk id).