mfs-find

Name: mfs-find
Author: zilliztech

29installs13forks

How to install mfs-find?

zilliztech/mfs/mfs-find

$npx -y skills add zilliztech/mfs --skill mfs-find

Installs into the current project.

Summary

Search, grep, browse, and read across registered MFS data sources via the mfs CLI — codebases, docs, PDFs, web crawls, databases (postgres/mysql/mongo/snowflake/bigquery), issue trackers (jira/linear/github), CRMs (hubspot), chat (slack/discord/gmail/feishu), object stores (s3/

Files · 1

View on GitHub

SKILL.md

1	# MFS — find / read across configured sources
2
3	## 1. What MFS is
4
5	A retrieval layer that exposes many kinds of content as a unified path
6	tree and makes that tree searchable through one hybrid index:
7
8	- One CLI (`mfs`), one mental model. Local dir, Postgres, GitHub repo,
9	Slack workspace, S3 bucket, BigQuery dataset — all addressed as paths
10	under their `<scheme>://` URI. Same verbs everywhere: `ls / tree / cat /
11	head / tail / grep / search / export`.
12	- One hybrid index. Dense vectors (semantic) + BM25 (keyword) fused
13	per query — covers conceptual recall and exact-token recall in one call.
14	- POSIX-style locators. Every search hit carries a `locator` that
15	reopens the exact unit: `{"lines":[s,e]}` for text/code, a PK dict for
16	rows/issues/threads.
17
18	## 2. When to use MFS — and when NOT to
19
20	\| Situation \| Use MFS? \|
21	\|---\|---\|
22	\| 1000+ files / rows / pages, you don't know where the answer is \| ✅ \|
23	\| Cross-source question ("any past tickets / commits / RFCs about X") \| ✅ `--all` \|
24	\| Concept-style query that won't match literally \| ✅ `--mode semantic` \|
25	\| You already know the exact file + roughly where to look \| ❌ plain `cat`/`grep` \|
26	\| Exact identifier / error code in 5 files you can list \| ❌ plain `grep`/`rg` \|
27	\| Real-time tailing of a live log \| ❌ index lags ingest \|
28	\| The source isn't in MFS yet \| wrong skill, use `mfs-ingest` to register first \|
29
30	Rule: use the smallest tool that answers the question. MFS pays off
31	when the scope is too big for `rg`.
32
33	Borderline — ASK the user:
34
35	\| Ask \| Likely answer \| Why \|
36	\|---\|---\|---\|
37	\| "Summarise these 10 PDFs" \| ✅ `mfs search` + `cat --peek` per hit \| each PDF gets a `converted_md` artifact + searchable chunks \|
38	\| "Find similar tickets to this one" \| ✅ paste the ticket text as the search query \| semantic over `row_text` does similarity matching \|
39	\| "Watch for new slack messages" \| ❌ no `watch` capability; use Slack's API \| index lags ingest \|
40	\| "Look up user 12345" \| ❌ `mfs cat <source> --locator '{"id":12345}'` directly (skip search) \| one-record-by-id doesn't need ranking \|
41
42	## 3. Pre-flight — confirm the source is indexed
43
44	Before running any query, especially on cross-source asks:
45
46	```bash
47	mfs status # server up? any connectors registered?
48	mfs connector inspect <uri> # this connector's object/job summary
49	mfs ls <uri> --json # per-entry capabilities + indexable / search_status
50	```
51
52	- Server unreachable → tell user to start it (`mfs serve start` if
53	self-hosted), or this skill can't proceed.
54	- `connectors` empty → user hasn't ingested anything yet. **Redirect to
55	`mfs-ingest`** — don't try to search nothing.
56	- `search_status: unavailable` for the target URI → only `grep` / `ls` /
57	`cat` work; offer those or redirect to `mfs-ingest` for a re-sync.
58	- `building` → sync in flight; fall back to `mfs grep` (works without an
59	index) until done.
60	- `partial` → recall incomplete but usable; flag the caveat to the user.
61
62	## 4. The core workflow: search → locate → browse
63
64	```
65	search locate browse
66	┌──────────────────┐ ┌──────────────────┐ ┌─────────────────────┐
67	│ semantic + BM25 │ → │ result has lines │ → │ cat --range / cat │
68	│ finds candidates│ │ or a locator │ │ --peek to confirm │
69	└──────────────────┘ └──────────────────┘ └─────────────────────┘
70	```
71
72	On large corpora this loop is the whole point: read only the part that
73	matters. On small corpora it's still fine, just lighter.
74
75	Concrete:
76
77	1. Search:
78	```bash
79	mfs search "<what the user actually wants>" <path-or-uri> --top-k 10
80	```
81	2. Locate — every hit's envelope carries `locator`:
82	- text/code → `{"lines":[start,end]}` → `mfs cat <source> --range start:end`
83	- structured (row/issue/thread) → PK dict → `mfs cat <source> --locator '{...}'`
84	- once-per-object (dir/schema summary, image VLM) → `null` → `mfs cat <source>`
85	3. Browse — verify only what's needed:
86	```bash
87	mfs cat --peek <file> # outline (headings / function signatures)
88	mfs cat --ski

People who liked this also viewed

6 picks

Type

mfs-find

29installs13forks

View on GitHub

How to install mfs-find?

zilliztech/mfs/mfs-find

$npx -y skills add zilliztech/mfs --skill mfs-find

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# MFS — find / read across configured sources
2
3	## 1. What MFS is
4
5	A retrieval layer that exposes many kinds of content as a unified path
6	tree and makes that tree searchable through one hybrid index:
7
8	- One CLI (`mfs`), one mental model. Local dir, Postgres, GitHub repo,
9	Slack workspace, S3 bucket, BigQuery dataset — all addressed as paths
10	under their `<scheme>://` URI. Same verbs everywhere: `ls / tree / cat /
11	head / tail / grep / search / export`.
12	- One hybrid index. Dense vectors (semantic) + BM25 (keyword) fused
13	per query — covers conceptual recall and exact-token recall in one call.
14	- POSIX-style locators. Every search hit carries a `locator` that
15	reopens the exact unit: `{"lines":[s,e]}` for text/code, a PK dict for
16	rows/issues/threads.
17
18	## 2. When to use MFS — and when NOT to
19
20	\| Situation \| Use MFS? \|
21	\|---\|---\|
22	\| 1000+ files / rows / pages, you don't know where the answer is \| ✅ \|
23	\| Cross-source question ("any past tickets / commits / RFCs about X") \| ✅ `--all` \|
24	\| Concept-style query that won't match literally \| ✅ `--mode semantic` \|
25	\| You already know the exact file + roughly where to look \| ❌ plain `cat`/`grep` \|
26	\| Exact identifier / error code in 5 files you can list \| ❌ plain `grep`/`rg` \|
27	\| Real-time tailing of a live log \| ❌ index lags ingest \|
28	\| The source isn't in MFS yet \| wrong skill, use `mfs-ingest` to register first \|
29
30	Rule: use the smallest tool that answers the question. MFS pays off
31	when the scope is too big for `rg`.
32
33	Borderline — ASK the user:
34
35	\| Ask \| Likely answer \| Why \|
36	\|---\|---\|---\|
37	\| "Summarise these 10 PDFs" \| ✅ `mfs search` + `cat --peek` per hit \| each PDF gets a `converted_md` artifact + searchable chunks \|
38	\| "Find similar tickets to this one" \| ✅ paste the ticket text as the search query \| semantic over `row_text` does similarity matching \|
39	\| "Watch for new slack messages" \| ❌ no `watch` capability; use Slack's API \| index lags ingest \|
40	\| "Look up user 12345" \| ❌ `mfs cat <source> --locator '{"id":12345}'` directly (skip search) \| one-record-by-id doesn't need ranking \|
41
42	## 3. Pre-flight — confirm the source is indexed
43
44	Before running any query, especially on cross-source asks:
45
46	```bash
47	mfs status # server up? any connectors registered?
48	mfs connector inspect <uri> # this connector's object/job summary
49	mfs ls <uri> --json # per-entry capabilities + indexable / search_status
50	```
51
52	- Server unreachable → tell user to start it (`mfs serve start` if
53	self-hosted), or this skill can't proceed.
54	- `connectors` empty → user hasn't ingested anything yet. **Redirect to
55	`mfs-ingest`** — don't try to search nothing.
56	- `search_status: unavailable` for the target URI → only `grep` / `ls` /
57	`cat` work; offer those or redirect to `mfs-ingest` for a re-sync.
58	- `building` → sync in flight; fall back to `mfs grep` (works without an
59	index) until done.
60	- `partial` → recall incomplete but usable; flag the caveat to the user.
61
62	## 4. The core workflow: search → locate → browse
63
64	```
65	search locate browse
66	┌──────────────────┐ ┌──────────────────┐ ┌─────────────────────┐
67	│ semantic + BM25 │ → │ result has lines │ → │ cat --range / cat │
68	│ finds candidates│ │ or a locator │ │ --peek to confirm │
69	└──────────────────┘ └──────────────────┘ └─────────────────────┘
70	```
71
72	On large corpora this loop is the whole point: read only the part that
73	matters. On small corpora it's still fine, just lighter.
74
75	Concrete:
76
77	1. Search:
78	```bash
79	mfs search "<what the user actually wants>" <path-or-uri> --top-k 10
80	```
81	2. Locate — every hit's envelope carries `locator`:
82	- text/code → `{"lines":[start,end]}` → `mfs cat <source> --range start:end`
83	- structured (row/issue/thread) → PK dict → `mfs cat <source> --locator '{...}'`
84	- once-per-object (dir/schema summary, image VLM) → `null` → `mfs cat <source>`
85	3. Browse — verify only what's needed:
86	```bash
87	mfs cat --peek <file> # outline (headings / function signatures)
88	mfs cat --ski

People who liked this also viewed

6 picks

Type