@getmarrow/mcp 3.9.25 → 3.9.26
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/README.md +8 -0
- package/package.json +1 -1
package/README.md
CHANGED
|
@@ -85,6 +85,14 @@ Full feature history, examples, and API reference live at [getmarrow.ai/docs](ht
|
|
|
85
85
|
|
|
86
86
|
---
|
|
87
87
|
|
|
88
|
+
## Human-Directed Attribution Status
|
|
89
|
+
|
|
90
|
+
The Marrow API supports privacy-preserving provenance fields on direct `POST /v1/decisions` calls: `source_kind`, `human_directed`, `source_confidence`, `instruction_ref`, `instruction_hash`, and `source_meta`. These fields classify instruction source class without identifying the human or storing raw prompts/PII.
|
|
91
|
+
|
|
92
|
+
Current MCP tools and passive hooks remain backward compatible and do **not** yet expose first-class provenance parameters or automatically mark prompts as `human_directed`. MCP provenance wiring is deferred; direct API users should follow the live API reference at https://getmarrow.ai/docs.
|
|
93
|
+
|
|
94
|
+
---
|
|
95
|
+
|
|
88
96
|
## Agent Value Report
|
|
89
97
|
|
|
90
98
|
`marrow_value_report` lets MCP agents pull owner-ready proof of Marrow value without a dashboard. The existing passive hooks remain the default install path: `PostToolUse` logs meaningful tool outcomes and `UserPromptSubmit` injects relevant context plus decision briefs for risky prompts.
|